A high-speed image processing analysis driven forging production process monitoring method

Abstract

The application relates to the technical field of industrial visual detection, and discloses a forging and pressing production process monitoring method driven by high-speed image processing and analysis, which comprises the following steps: extracting a mill scale candidate feature image set; constructing a mill scale flying track set; generating a mill scale flying area feature map; constructing a die cavity risk propagation model based on the mill scale flying area feature map; and monitoring the forging and pressing process according to the mill scale abnormal accumulation risk grade. Compared with the prior art, which mainly relies on manual inspection to monitor the mill scale falling condition in the forging and pressing production process, especially under the conditions of high-temperature strong radiation, high-speed mill scale flying and dense particle shielding, the technical problem that real-time evaluation of the mill scale flying behavior cannot be realized is solved. Since the mill scale track recognition mechanism based on high-speed image sequence analysis is constructed, continuous monitoring of the mill scale flying behavior in the forging and pressing production process is realized, and the real-time performance of the forging and pressing production process monitoring is improved.

Description

Technical Field

[0001] This invention relates to the field of industrial visual inspection technology, and in particular to a high-speed image processing and analysis-driven method for monitoring the forging production process. Background Technology

[0002] Currently, during hot forging, the billet's surface develops a certain thickness of oxide scale due to prolonged exposure to air at high temperatures. When the billet enters the die for forging, this oxide scale easily detaches under impact loads and friction, forming high-speed, scattered particles. If this detached oxide scale enters the die cavity, it may cause defects such as indentations, inclusions, or surface roughness on the forging surface, thus affecting the surface quality and yield of the finished product. Therefore, real-time monitoring of oxide scale detachment and timely assessment of its accumulation risk during forging are crucial for improving the stability of forging quality and the safety of the production process.

[0003] However, monitoring of oxide scale shedding during forging production mainly relies on manual inspections or simple image acquisition using ordinary industrial cameras followed by offline analysis. This approach remains significantly insufficient in actual production environments. For example, hot forging stations typically experience complex conditions such as strong high-temperature radiation, drastic changes in lighting, and high-speed dispersion of oxide scale particles. Traditional visual inspection methods are prone to image brightness drift or thermal noise interference under strong heat radiation, leading to inaccurate oxide scale particle identification. Furthermore, existing methods often focus on target recognition within single-frame images, making it difficult to analyze the continuous movement trajectory of oxide scale particles and determine whether dispersed particles might enter the die cavity and accumulate. Therefore, current technologies cannot adequately meet the needs for real-time dynamic monitoring of oxide scale dispersion and accurate assessment of die cavity accumulation risks under high-temperature forging conditions.

[0004] Therefore, there is an urgent need for a forging production process monitoring method that can still achieve real-time identification of oxide scale shedding behavior, dynamic analysis of dispersion trajectory, and intelligent assessment of die cavity accumulation risk under high-temperature radiation environment and high-speed particle dispersion conditions, so as to improve the accuracy and real-time performance of forging production process monitoring and further enhance the quality of forging products and production stability. Summary of the Invention

[0005] To address the aforementioned technical shortcomings, the present invention aims to propose a high-speed image processing and analysis-driven monitoring method for forging production processes. This method addresses the technical problem that existing technologies primarily rely on manual inspection to monitor oxide scale shedding during forging production, particularly in high-temperature and high-radiation environments, high-speed oxide scale dispersion, and dense particle obstruction conditions, where real-time assessment of oxide scale dispersion behavior is impossible.

[0006] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a high-speed image processing analysis-driven method for monitoring the forging production process.

[0007] The high-speed image processing and analysis-driven forging production process monitoring method includes: Step S10: Obtain high-speed image sequence data and mold motion state data at the hot forging station in the forging production line. Based on the high-speed image sequence data and mold motion state data, use a high-temperature visual feature construction mechanism based on temperature radiation compensation to perform the task of extracting significant oxide scale features and output a set of candidate oxide scale feature images. Step S20: Based on the set of candidate feature images of oxide scale, a scattering behavior recognition mechanism based on multi-scale particle trajectory evolution is used to perform the task of analyzing the motion trajectory of oxide scale particles, and output the set of oxide scale scattering trajectories; Step S30: Based on the set of oxide scale scattering trajectories, a dynamic photothermal noise suppression mechanism is used to perform the abnormal particle region enhancement task, and the oxide scale scattering region feature map is output; Step S40: Based on the feature map of the oxide scale scattering area, an abnormal accumulation discrimination mechanism based on the cavity risk propagation model is used to perform the oxide scale accumulation risk assessment task and output the abnormal accumulation risk level of oxide scale; Step S50: Make monitoring decisions for the forging process based on the risk level of abnormal oxide scale accumulation, and output the monitoring results of forging production status.

[0008] Preferably, step S10, which involves acquiring high-speed image sequence data and die motion state data at the hot forging station in the forging production line, and performing the task of extracting significant oxide scale features using a high-temperature visual feature construction mechanism based on temperature radiation compensation, and outputting a set of candidate oxide scale feature images, specifically includes: Step S101: Acquire high-speed image sequence data and die motion status data at the hot forging station in the forging production line. The die motion status data includes forging stroke position, slider speed and impact moment signal. Step S102: Perform high-temperature radiance compensation processing on the high-speed image sequence data and construct a thermal radiation compensation function. , ;in, x is the pixel x pixel coordinate The original brightness value at time t. x is the pixel x pixel coordinate The compensated brightness value at time t This is the temperature radiation compensation coefficient. For a moment Estimated surface temperature of the billet; Step S103: Based on the thermal radiation compensation function The local brightness change rate was calculated using local window difference analysis. With spatial gradient strength And based on the local brightness change rate With spatial gradient strength Constructing a significant feature index of oxide skin ; Step S104: When the oxide scale significant characteristic index Greater than the preset significance threshold for oxide scale When the corresponding pixel position belongs to the oxide scale candidate region, the set of oxide scale candidate feature images is output.

[0009] Preferably, in step S20, the step of performing oxide scale particle motion trajectory analysis based on the oxide scale candidate feature image set and using a multi-scale particle trajectory evolution-based scattering behavior recognition mechanism to output the oxide scale scattering trajectory set specifically includes: Step S201: Based on the candidate feature image set of oxide scale, multi-scale feature point detection and sparse optical flow tracking feature extraction are performed using the OpenCV computer vision processing library and NumPy numerical computation library in Python to obtain oxide scale particle features. The oxide scale particle features include the horizontal displacement of oxide scale particles. Vertical displacement of oxide scale particles ; Step S202: Based on the horizontal displacement of the oxide scale particles Vertical displacement of oxide scale particles Calculate the velocity of the particles ; Step S203: Based on the particle movement speed A set of oxide scale scattering trajectories is constructed and output using a fusion method based on time series trajectory association and Kalman filter trajectory prediction.

[0010] Preferably, in step S202, ,in, The time interval between adjacent image frames in the set of candidate feature images for oxide scale is denoted as .

[0011] Preferably, step S30, which involves performing an abnormal particle region enhancement task based on the oxide scale scattering trajectory set using a dynamic photothermal noise suppression mechanism and outputting an oxide scale scattering region feature map, specifically includes: Step S301: Based on the set of oxide scale scattering trajectories, use trajectory spatial clustering analysis to perform particle trajectory endpoint clustering and output the scattering and landing areas; Step S302: Calculate the standard deviation of the brightness of the fixed-fall pixels in the scattered fixed-fall area and the average brightness of the fixed-fall pixels, and construct the thermal noise discrimination coefficient based on the standard deviation of the brightness of the fixed-fall pixels and the average brightness of the fixed-fall pixels. ; Step S303: When the thermal noise discrimination coefficient Exceeding the preset thermal noise discrimination threshold When the corresponding pixel is determined to be a thermal radiation noise pixel, a time-series filtering process is performed on the thermal radiation noise pixel, and finally the oxide scale scattering area feature map is output.

[0012] Preferably, step S40, which involves performing an abnormal accumulation risk assessment task based on the abnormal accumulation discrimination mechanism of the mold cavity risk propagation model using the feature map of the oxide scale scattering area, and outputting the abnormal accumulation risk level of the oxide scale, specifically includes: Step S401: Based on the feature map of the oxide scale scattering area, the kernel density estimation method is used to estimate the probability of particles falling into the transverse cavity region. ; Step S402: Based on the feature map of the oxide scale scattering area, use connected component statistical analysis to count the number of oxide scale particles in the transverse cavity area. ; Step S403: Based on the probability of particles falling into the transverse cavity region and the number of oxide scale particles in the transverse cavity region The Sigmoid risk mapping function is used for nonlinear fusion mapping to output the mold cavity contamination risk index. Based on the mold cavity contamination risk index, the risk level of abnormal oxide scale accumulation is set and output.

[0013] Preferably, step S50, which involves making monitoring decisions for the forging process based on the risk level of abnormal oxide scale accumulation and outputting the monitoring results of the forging production status, specifically includes: Step S501: Obtain particle movement direction information from the oxide scale scattering trajectory set, and perform weighted linear fusion based on the particle movement direction information to assess the risk level of abnormal oxide scale accumulation, and output the comprehensive evaluation index of forging process risk. ; Step S502: Calculate the comprehensive risk assessment index for the forging process. Compared with the preset first risk threshold Second risk threshold The comparison is used to classify and determine the forging production status, and output a safety warning status, specifically including: when When the safety status warning state is determined to be the normal operating state; when When the safety warning status is changed to a risk warning status, the warning status is determined to be changed to a risk warning status. when When the safety status warning is issued, it is determined to be an abnormal and dangerous state; Step S503: When the safety status warning state is determined to be a risk warning state or an abnormal dangerous state, a forging process warning signal is generated, and the forging production status monitoring result is output.

[0014] This invention also provides a high-speed image processing and analysis-driven forging production process monitoring system, comprising: The high-temperature visual feature construction module is used to acquire high-speed image sequence data and mold motion state data at the hot forging station in the forging production line. Based on the high-speed image sequence data and mold motion state data, a high-temperature visual feature construction mechanism based on temperature radiation compensation is used to perform the task of extracting significant oxide scale features and output a set of candidate oxide scale feature images. The particle trajectory recognition module is used to perform the task of analyzing the motion trajectory of oxide scale particles based on the oxide scale candidate feature image set and adopts a scattering behavior recognition mechanism based on multi-scale particle trajectory evolution, and outputs a set of oxide scale scattering trajectories. The photothermal noise suppression module is used to perform abnormal particle region enhancement tasks based on the set of oxide scale scattering trajectories using a dynamic photothermal noise suppression mechanism, and outputs an oxide scale scattering region feature map. The mold cavity risk assessment module is used to perform an abnormal accumulation risk assessment task based on the abnormal accumulation discrimination mechanism based on the mold cavity risk propagation model, based on the feature map of the oxide scale scattering area, and output the abnormal accumulation risk level of oxide scale. The production monitoring module is used to make monitoring decisions for the forging process based on the risk level of abnormal oxide scale accumulation, and output the monitoring results of the forging production status.

[0015] The present invention also provides a high-speed image processing analysis-driven forging production process monitoring device, comprising: a memory, a processor, and a high-speed image processing analysis-driven forging production process monitoring program stored in the memory and executable on the processor. When the high-speed image processing analysis-driven forging production process monitoring program is executed by the processor, a high-speed image processing analysis-driven forging production process monitoring method is implemented.

[0016] The present invention also provides a computer program product, including a high-speed image processing analysis driven forging production process monitoring program, wherein the high-speed image processing analysis driven forging production process monitoring program implements the high-speed image processing analysis driven forging production process monitoring method when executed by a processor.

[0017] The beneficial effects of this invention are as follows: By constructing a high-temperature visual feature extraction mechanism based on temperature radiation compensation and an oxide scale scattering behavior recognition mechanism based on multi-scale particle trajectory evolution, this invention continuously and dynamically analyzes the scattering trajectory of oxide scale particles during the forging process. Compared with the existing technology that mainly relies on single-frame image detection or manual observation, this invention can accurately identify the shedding and scattering behavior of oxide scale under high temperature, strong radiation and high-speed particle movement environments, thereby significantly improving the stability and real-time performance of oxide scale detection.

[0018] This invention establishes a cavity risk propagation model based on the feature map of the scattering area, and constructs a scale accumulation risk assessment mechanism by combining the probability of particles falling into the cavity area and the number of particles in the area. This enables quantitative analysis and graded early warning of the risk of abnormal scale accumulation. Compared with the traditional method of judging based on experience or simple threshold monitoring, it can more accurately predict the risk of scale entering the cavity and causing defects on the surface of forgings, thereby improving the intelligence level of monitoring the forging production process and effectively improving the quality of forging products. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart illustrating the first embodiment of a high-speed image processing and analysis-driven forging production process monitoring method of the present invention.

[0021] Figure 2 This is a schematic diagram of the equipment for a high-speed image processing and analysis-driven monitoring method for forging production processes according to the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the high-speed image processing analysis-driven forging production process monitoring method of the present invention, which proposes the first embodiment of the high-speed image processing analysis-driven forging production process monitoring method of the present invention.

[0024] In the first embodiment, the high-speed image processing analysis-driven forging production process monitoring method includes: Step S10: Obtain high-speed image sequence data and mold motion state data at the hot forging station in the forging production line. Based on the high-speed image sequence data and mold motion state data, use a high-temperature visual feature construction mechanism based on temperature radiation compensation to perform the task of extracting significant oxide scale features and output a set of candidate oxide scale feature images. It should be noted that the "high-temperature visual feature construction mechanism based on temperature radiation compensation" refers to a visual feature construction method that addresses the strong thermal radiation background generated by the high temperature of the billet surface in the hot forging station environment. This is achieved by establishing a temperature radiation compensation model to dynamically correct pixel brightness in high-speed images, and by combining local brightness change features with spatial texture gradient features to perform saliency analysis on oxide scale areas. Specifically, this mechanism mainly includes the following aspects: First, a continuous high-speed image sequence is acquired at the hot forging station using a high-speed industrial camera, and the image acquisition time is synchronously calibrated by combining the mold motion state data, ensuring that the image information and the forging motion state remain synchronized in the time dimension. First, the original image brightness is corrected using a temperature radiation compensation model to reduce the impact of high-temperature radiation from the billet on the image's grayscale distribution. Second, pixel brightness variation features are calculated using local window analysis in the compensated image, and image texture structure features are extracted using spatial gradient analysis. Finally, a significant oxide scale feature index is constructed by fusing brightness variation features and spatial gradient features, thereby identifying and marking the locations of potential oxide scale flaking areas in the image. Through these processes, a stable visual feature representation can be constructed in a complex high-temperature environment, making the structural features of the oxide scale region in the image clearer, thus forming a set of candidate oxide scale feature images.

[0025] Understandably, introducing a temperature radiation compensation mechanism during the image feature extraction stage can effectively reduce the overall offset effect of high-temperature radiation environment on image brightness distribution, correcting the brightness drift caused by thermal radiation and thus enhancing the visual contrast between the oxide scale area and the billet metal matrix. Simultaneously, through joint analysis of local brightness change rate and spatial gradient intensity, regions with obvious structural changes can be identified in complex backgrounds. Since oxide scale typically forms regions with distinct boundaries and texture differences during detachment or peeling, constructing an oxide scale saliency index can more stably extract oxide scale candidate regions. The oxide scale candidate feature image set obtained through this step not only contains spatial location information of the oxide scale region but also provides a reliable initial visual feature basis for oxide scale particle trajectory identification in subsequent steps, thereby improving the accuracy and stability of subsequent scattering trajectory analysis.

[0026] It should be understood that, compared to traditional forging production monitoring methods that directly perform fixed threshold segmentation or simple edge detection on the original image, this invention, by introducing a temperature radiation compensation model and a multi-feature fusion saliency analysis mechanism in the image processing stage, can effectively reduce the impact of high-temperature radiation environment on image brightness distribution, and improve the stability of oxide scale region identification through comprehensive analysis of brightness change features and texture gradient features. Traditional methods in high-temperature forging environments are often easily affected by factors such as changes in illumination, thermal radiation interference, and metal surface reflection, leading to unstable oxide scale region identification or a high false detection rate. This invention, by establishing a visual feature construction mechanism combining temperature radiation compensation and saliency index, can maintain good recognition performance even under complex high-temperature backgrounds, making the extraction of oxide scale candidate regions more reliable and providing a stable data foundation for subsequent oxide scale scattering trajectory analysis and mold cavity risk assessment. For example, in actual forging production environments, when hot forging billets enter the mold area for forming processing, the billet surface is usually accompanied by significant thermal radiation, causing a shift in the overall brightness of the image obtained by the image acquisition device. In this case, if region identification is performed directly on the original image, some bright areas are easily misidentified as oxide scale regions. By employing the temperature radiation compensation and salient feature extraction mechanism described in this invention, image brightness can first be radiatively compensated to restore it to a state closer to the true surface reflection characteristics. Then, through comprehensive analysis of local brightness changes and spatial texture changes, regions with significant structural differences are identified, thereby accurately extracting oxide scale candidate regions. Multiple tests in actual production environments have shown that, under complex high-temperature background conditions, the method described in this invention can more stably identify oxide scale regions and effectively reduce misidentification caused by thermal radiation or surface reflection, thus providing a reliable image feature basis for subsequent oxide scale scattering behavior analysis and forging production process monitoring.

[0027] Step S20: Based on the set of candidate feature images of oxide scale, a scattering behavior recognition mechanism based on multi-scale particle trajectory evolution is used to perform the task of analyzing the motion trajectory of oxide scale particles, and output the set of oxide scale scattering trajectories; It should be noted that the "scattering behavior recognition mechanism based on multi-scale particle trajectory evolution" in this step refers to a dynamic analysis method for constructing the motion trajectory of oxide scale particles in the time dimension by performing multi-scale analysis on the particle position changes in a continuous high-speed image sequence, targeting the motion characteristics of oxide scale particles that detach and form high-speed scattered particles under forging and impact. This mechanism mainly includes the following aspects: First, particle regions with obvious oxide scale characteristics are identified in the oxide scale candidate feature image set obtained in step S10, and feature point matching is performed on these particle regions between adjacent image frames to obtain the spatial position change information of particles in continuous images; Second, by analyzing the position changes of particles between different time frames, the displacement changes of particles in the image plane are calculated to obtain the motion direction and velocity of particles; Third, by comprehensively analyzing the particle motion information at different spatial scales, the motion behavior characteristics of oxide scale particles of different sizes during the scattering process can be identified; Finally, the particle motion states in continuous frames are associated through a time series trajectory association method to form a complete oxide scale particle scattering trajectory and output an oxide scale scattering trajectory set. This mechanism allows for a continuous description of the movement of oxide scale particles under forging and impact, thus providing an important data foundation for subsequent oxide scale accumulation risk assessment.

[0028] Understandably, analyzing the motion trajectory of oxide scale particles in high-speed image sequences can clearly reflect the scattering behavior characteristics of oxide scale during forging impact, such as the scattering direction, scattering range, and changes in scattering velocity. Since oxide scale particles typically undergo high-speed motion at the moment of mold impact, relying solely on single-frame images makes it difficult to accurately determine whether particles are in a scattered state or whether they might enter the mold cavity. However, by constructing continuous particle motion trajectories, the particle motion process can be tracked over time, thus more accurately describing oxide scale scattering behavior. This trajectory evolution-based analysis method can effectively distinguish between static oxide scale residue areas and dynamic scattered particle areas, providing a reliable basis for subsequent scattering area identification and risk assessment.

[0029] It should be understood that, compared to traditional forging production monitoring technologies that identify oxide scale areas through simple image segmentation or static target detection, this invention, by performing trajectory-level analysis of particle motion in continuous image frames, can characterize the motion characteristics of oxide scale particles in the temporal dimension. It can not only identify the spatial location of oxide scale particles but also their motion trend and dispersion path. Traditional methods typically only identify the presence of oxide scale but struggle to determine whether it disperses during forging and whether the dispersion direction points towards the die cavity area, thus hindering effective assessment of potential quality risks. This invention, by constructing a set of oxide scale particle motion trajectories, can more accurately reflect oxide scale dispersion behavior, thereby providing crucial input information for subsequent die cavity risk propagation analysis and improving the intelligence level of forging production process monitoring.

[0030] For example, in actual forging production, when the billet enters the die area and is subjected to impact loads, some oxide scale will peel off from the billet surface and scatter in a certain direction. In high-speed image sequences, these oxide scale particles typically exhibit significant positional changes between consecutive image frames. By continuously tracking the particle regions in the oxide scale candidate feature image set, it can be observed that some particles move rapidly along the die opening direction, while others diffuse towards the outer area of ​​the die. By performing correlation analysis on the motion trajectories of these particles, a complete set of scattering trajectories can be formed, and it can be further determined whether the particle scattering direction spatially intersects with the die cavity area. In this way, the scattering behavior of oxide scale particles can be identified in real time during the forging process, thus providing a reliable trajectory information basis for subsequent oxide scale accumulation risk analysis and production process monitoring.

[0031] Step S30: Based on the set of oxide scale scattering trajectories, a dynamic photothermal noise suppression mechanism is used to perform the abnormal particle region enhancement task, and the oxide scale scattering region feature map is output; It should be noted that the "dynamic photothermal noise suppression mechanism" in this step refers to an image enhancement method that dynamically suppresses image noise generated in the hot forging production environment due to high-temperature radiation, metal surface reflection, and changes in ambient light. This is achieved by combining information on oxide scale scattering trajectories to identify abnormal brightness areas in the image. The method includes: determining the spatial regions where oxide scale particles may appear based on the oxide scale scattering trajectory set obtained in step S20; performing cluster analysis on the trajectory endpoints and their neighboring regions to identify the main landing areas of the oxide scale particles; statistically analyzing the pixel brightness distribution within these areas after determining the landing areas, and identifying abnormal brightness noise caused by high-temperature radiation or metal reflection by calculating the pixel brightness change characteristics; subsequently performing temporal filtering and spatial smoothing on the identified thermal radiation noise pixels to reduce the interference of thermal radiation noise on the image structure while maintaining the structural characteristics of the oxide scale particles; and finally, enhancing the abnormal particle areas to give the oxide scale scattering areas clearer structural boundaries in the image, thus forming an oxide scale scattering area feature map.

[0032] Understandably, due to the high temperatures in the hot forging production environment, the billet surface typically produces noticeable thermal radiation spots. Simultaneously, the die surface and metal surface may also generate localized reflections. These factors all contribute to areas of abnormal brightness in the image. If the original image is directly used for oxide scale area analysis, these photothermal noises are easily misidentified as oxide scale particles, affecting the accuracy of subsequent trajectory determination and risk assessment. By introducing a dynamic photothermal noise suppression mechanism in this step, targeted noise filtering of the image area can be performed based on oxide scale scattering trajectory information. This strengthens the actual oxide scale particle areas while effectively suppressing noise areas generated by thermal radiation or reflection, making the oxide scale scattering areas in the image clearer and more stable. This improves the data reliability in the subsequent die cavity risk assessment process. For example, in actual forging production, when the billet is impacted in the die, some oxide scale will detach from the billet surface and scatter in a certain direction. In high-speed image sequences, these particles often exhibit certain trajectory distribution characteristics in space, while the thermal radiation bright spots in the image typically appear as areas with sudden increases in brightness but lacking continuous motion trajectories. By analyzing the set of oxide scale scattering trajectories obtained in step S20, the spatial regions where oxide scale particles are mainly distributed can be determined. Then, statistical analysis of image brightness changes within these regions identifies the actual oxide scale areas consistent with the particle movement trajectories and suppresses abnormal brightness areas caused by thermal radiation or reflection. After this processing, the boundaries of the oxide scale scattering areas in the image are clearer, and the particle structure features are more obvious, thus providing more reliable image feature information for subsequent oxide scale accumulation risk assessment.

[0033] Step S40: Based on the feature map of the oxide scale scattering area, an abnormal accumulation discrimination mechanism based on the cavity risk propagation model is used to perform the oxide scale accumulation risk assessment task and output the abnormal accumulation risk level of oxide scale; It should be noted that the "abnormal accumulation discrimination mechanism based on the mold cavity risk propagation model" in this step refers to an analytical method that, after obtaining the feature map of the oxide scale scattering area, analyzes the spatial distribution characteristics of oxide scale particles, the endpoint position of the scattering trajectory, and the spatial aggregation of particles in the mold structure area to assess the risk of oxide scale particles entering the mold cavity area and forming accumulation. This mechanism mainly includes the following aspects: First, the spatial distribution area of ​​oxide scale particles is identified based on the feature map of the oxide scale scattering area, and the possible landing point area of ​​the particles is determined by combining the scattering trajectory information obtained in step S20; second, the mold structure area is divided into multiple functional areas, including the mold cavity area, the mold surface area, and the mold periphery area, and the probability of particles entering the mold cavity area is estimated by analyzing the spatial relationship between the particle landing point and the mold cavity area; third, a mold cavity contamination risk index is constructed by statistically analyzing the particle distribution density within the mold cavity area and the degree of particle aggregation near the boundary of the mold cavity area; finally, the oxide scale accumulation risk is graded and assessed based on the mold cavity contamination risk index, thereby outputting the abnormal accumulation risk level of oxide scale. This mechanism allows for a comprehensive assessment of the risk of oxide scale buildup in the mold area from two dimensions: spatial distribution and particle aggregation.

[0034] Understandably, during the forging process, when oxide scale particles scatter under impact loads, their trajectory and impact point distribution are often closely related to the die structure. If a large number of oxide scale particles enter the die cavity, they may be pressed into the forging surface during subsequent forging processes, forming indentations or surface defects. Therefore, by spatially analyzing the characteristic map of the oxide scale scattering area and constructing a die cavity risk propagation model based on particle scattering trajectory information, the propagation path of oxide scale particles in the die area and possible accumulation areas can be predicted, thereby identifying areas where abnormal accumulation may occur in advance. In this way, risk assessment can be conducted before severe accumulation of oxide scale particles, providing an important basis for subsequent production monitoring decisions.

[0035] It should be understood that, compared to traditional forging production monitoring technologies that rely solely on manual experience or simple image detection to identify oxide scale accumulation, this invention, by constructing a cavity risk propagation model and combining it with spatial analysis of oxide scale scattering area feature maps, can more accurately assess the distribution trend of oxide scale particles within the die structure area. Traditional methods often only detect oxide scale problems after surface defects have formed on the forging, making it difficult to identify potential risks in advance during production. This invention, however, by performing trajectory analysis on oxide scale scattering behavior and further evaluating particle landing point distribution through the cavity risk propagation model, can identify oxide scale accumulation risks in real time during production, thereby providing early warning information to the production control system and preventing the degradation of forging surface quality due to oxide scale entering the die cavity. For example, in an actual forging production environment, when a billet enters the die area and is impacted, some oxide scale will peel off from the billet surface and scatter along the die opening direction. In high-speed image sequences, the aforementioned steps can identify the scattering area of ​​oxide scale particles and obtain the spatial distribution information of the particles. When the scattering trajectories of these particles gradually converge towards the mold cavity area, the mold cavity risk propagation model can identify a particle aggregation trend in this area and determine the potential risk of oxide scale accumulation. During production, when oxide scale particles are detected to exhibit a significant aggregation trend near the mold cavity area, it can be determined that there is a potential accumulation risk in the current forging process. When the risk level reaches a preset condition, the corresponding risk level information is output, thus providing a basis for subsequent production monitoring and process adjustments.

[0036] Step S50: Make monitoring decisions for the forging process based on the risk level of abnormal oxide scale accumulation, and output the monitoring results of forging production status.

[0037] It should be noted that the "forging process monitoring and decision-making" in this step refers to a production control decision-making process that, after obtaining the risk level of abnormal oxide scale accumulation, comprehensively analyzes the current forging production status, judges the operating status of production equipment, the execution status of the process, and potential quality risks, and outputs corresponding production status monitoring results. This monitoring and decision-making process includes: assessing the risk level of the current forging production status based on the abnormal oxide scale accumulation risk level obtained in step S40; determining whether there are risks that may affect the surface quality of the forgings during the current forging process, combined with die movement data and the distribution of oxide scale scattering areas; and generating corresponding production monitoring information and outputting forging production status monitoring results when the abnormal accumulation risk is identified to meet preset conditions. The monitoring results may include the current forging production status category, risk warning information, and corresponding production status prompts, thus providing reference monitoring results for the production management system or operators.

[0038] Understandably, by introducing the risk level of abnormal oxide scale accumulation as a basis for monitoring decisions in the forging production monitoring system, the production monitoring process can be transformed from a traditional passive detection mode to a proactive identification and risk warning mode. When a low risk of oxide scale accumulation is identified, the current production process can be determined to be in a normal state; when an accumulation trend of oxide scale particles is identified near the die cavity area, the current production process can be determined to have potential risks, and corresponding risk warning information will be output; when an obvious accumulation trend of oxide scale particles is identified in the die cavity area, the current production process can be determined to potentially affect the surface quality of the forging, and the production system will be prompted to perform necessary production status monitoring or handling operations. In this way, the production monitoring system can identify potential quality risks in a timely manner during the forging process, thereby improving the safety and stability of the production process.

[0039] It should be understood that, compared to traditional forging processes that rely primarily on manual experience or post-forging quality inspection, this invention, by analyzing oxide scale dispersion behavior in real time and constructing an oxide scale accumulation risk assessment model during production, can identify potential quality risks in advance during the forging process. Traditional methods often only detect surface defects after forging is completed through manual inspection or offline testing, by which time some production loss has already occurred. This invention, however, by analyzing oxide scale dispersion trajectories and accumulation risks in real time, can identify risks in advance during production and output production monitoring results. This allows the production management system to promptly grasp the current forging status, reduce surface quality problems caused by oxide scale accumulation, and improve the controllability and stability of the forging production process.

[0040] Example 2: Furthermore, the high-speed image processing analysis-driven forging production process monitoring system provided by this invention employs a high-speed image processing analysis-driven forging production process monitoring method from the above embodiments, which can solve the technical problem of high-speed image processing analysis-driven forging production process monitoring. The beneficial effects of the high-speed image processing analysis-driven forging production process monitoring system provided by this invention are the same as those of the high-speed image processing analysis-driven forging production process monitoring method provided in the above embodiments, and other technical features of the high-speed image processing analysis-driven forging production process monitoring system are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0041] Example 3: This invention provides a high-speed image processing and analysis-driven monitoring device for forging production processes. Please refer to... Figure 2A high-speed image processing analysis-driven forging production process monitoring device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, enable the at least one processor to execute the high-speed image processing analysis-driven forging production process monitoring method described in Embodiment 1 above. The high-speed image processing analysis-driven forging production process monitoring device in this embodiment may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This high-speed image processing analysis-driven forging production process monitoring device is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this invention. A high-speed image processing and analysis-driven forging production process monitoring device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the high-speed image processing and analysis-driven forging production process monitoring device. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An I / O interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows a high-speed image processing analysis-driven forging process monitoring device to communicate wirelessly or wiredly with other devices to exchange data. While the figure shows a high-speed image processing analysis-driven forging process monitoring device with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0042] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the high-speed image processing analysis-driven forging production process monitoring method described above. The computer program product provided by this invention can solve the technical problem of high-speed image processing analysis-driven forging production process monitoring. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the high-speed image processing analysis-driven forging production process monitoring method provided in the above embodiments, and will not be repeated here.

[0043] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0044] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0045] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A high-speed image processing and analysis-driven method for monitoring the forging production process, characterized in that, The methods include: Step S10: Obtain high-speed image sequence data and mold motion state data at the hot forging station in the forging production line. Based on the high-speed image sequence data and mold motion state data, use a high-temperature visual feature construction mechanism based on temperature radiation compensation to perform the task of extracting significant oxide scale features and output a set of candidate oxide scale feature images. Step S20: Based on the set of candidate feature images of oxide scale, a scattering behavior recognition mechanism based on multi-scale particle trajectory evolution is used to perform the task of analyzing the motion trajectory of oxide scale particles, and output the set of oxide scale scattering trajectories; Step S30: Based on the set of oxide scale scattering trajectories, a dynamic photothermal noise suppression mechanism is used to perform the abnormal particle region enhancement task, and the oxide scale scattering region feature map is output; Step S40: Based on the feature map of the oxide scale scattering area, an abnormal accumulation discrimination mechanism based on the cavity risk propagation model is used to perform the oxide scale accumulation risk assessment task and output the abnormal accumulation risk level of oxide scale; Step S50: Make monitoring decisions for the forging process based on the risk level of abnormal oxide scale accumulation, and output the monitoring results of forging production status.

2. The high-speed image processing analysis-driven forging production process monitoring method as described in claim 1, characterized in that, Step S10 specifically includes: Step S101: Acquire high-speed image sequence data and die motion status data at the hot forging station in the forging production line. The die motion status data includes forging stroke position, slider speed and impact moment signal. Step S102: Perform high-temperature radiance compensation processing on the high-speed image sequence data and construct a thermal radiation compensation function. , ;in, x is the pixel x pixel coordinate The original brightness value at time t. x is the pixel x pixel coordinate The compensated brightness value at time t This is the temperature radiation compensation coefficient. For a moment Estimated surface temperature of the billet; Step S103: Based on the thermal radiation compensation function The local brightness change rate was calculated using local window difference analysis. With spatial gradient strength And based on the local brightness change rate With spatial gradient strength Constructing a significant feature index of oxide skin ; Step S104: When the oxide scale significant characteristic index Greater than the preset oxide scale significance threshold When the corresponding pixel position belongs to the oxide scale candidate region, the set of oxide scale candidate feature images is output.

3. The high-speed image processing analysis-driven forging production process monitoring method as described in claim 1, characterized in that, Step S20 specifically includes: Step S201: Based on the candidate feature image set of oxide scale, multi-scale feature point detection and sparse optical flow tracking feature extraction are performed using the OpenCV computer vision processing library and NumPy numerical computation library in Python to obtain oxide scale particle features. The oxide scale particle features include the horizontal displacement of oxide scale particles. Vertical displacement of oxide scale particles ; Step S202: Based on the horizontal displacement of the oxide scale particles Vertical displacement of oxide scale particles Calculate the velocity of the particles ; Step S203: Based on the particle movement speed A set of oxide scale scattering trajectories is constructed and output using a fusion method based on time series trajectory association and Kalman filter trajectory prediction.

4. The high-speed image processing analysis-driven forging production process monitoring method as described in claim 3, characterized in that, In step S202, ,in, The time interval between adjacent image frames in the set of candidate feature images for oxide scale is denoted as .

5. The high-speed image processing analysis-driven forging production process monitoring method as described in claim 1, characterized in that, Step S30 specifically includes: Step S301: Based on the set of oxide scale scattering trajectories, use trajectory spatial clustering analysis to perform particle trajectory endpoint clustering and output the scattering and landing areas; Step S302: Calculate the standard deviation of the brightness of the fixed-fall pixels in the scattered fixed-fall area and the average brightness of the fixed-fall pixels, and construct the thermal noise discrimination coefficient based on the standard deviation of the brightness of the fixed-fall pixels and the average brightness of the fixed-fall pixels. ; Step S303: When the thermal noise discrimination coefficient Exceeding the preset thermal noise discrimination threshold When the corresponding pixel is determined to be a thermal radiation noise pixel, a time-series filtering process is performed on the thermal radiation noise pixel, and finally the oxide scale scattering area feature map is output.

6. The high-speed image processing analysis-driven forging production process monitoring method as described in claim 1, characterized in that, Step S40 specifically includes: Step S401: Based on the feature map of the oxide scale scattering area, the kernel density estimation method is used to estimate the probability of particles falling into the transverse cavity region. ; Step S402: Based on the feature map of the oxide scale scattering area, use connected component statistical analysis to count the number of oxide scale particles in the transverse cavity area. ; Step S403: Based on the probability of particles falling into the transverse cavity region and the number of oxide scale particles in the transverse cavity region The Sigmoid risk mapping function is used for nonlinear fusion mapping to output the mold cavity contamination risk index. Based on the mold cavity contamination risk index, the risk level of abnormal oxide scale accumulation is set and output.

7. The high-speed image processing analysis-driven forging production process monitoring method as described in claim 1, characterized in that, Step S50 specifically includes: Step S501: Obtain particle movement direction information from the oxide scale scattering trajectory set, and perform weighted linear fusion based on the particle movement direction information to assess the risk level of abnormal oxide scale accumulation, and output the comprehensive evaluation index of forging process risk. ; Step S502: Calculate the comprehensive risk assessment index for the forging process. Compared with the preset first risk threshold Second risk threshold The comparison is used to classify and determine the forging production status, and output a safety warning status, specifically including: when When the safety status warning state is determined to be the normal operating state; when When the safety warning status is changed to a risk warning status, the warning status is determined to be changed to a risk warning status. when When the safety status warning is issued, it is determined to be an abnormal and dangerous state; Step S503: When the safety status warning state is determined to be a risk warning state or an abnormal dangerous state, a forging process warning signal is generated, and the forging production status monitoring result is output.

8. A high-speed image processing analysis-driven forging production process monitoring system, applied to the high-speed image processing analysis-driven forging production process monitoring method according to any one of claims 1 to 7, characterized in that, The high-speed image processing and analysis-driven forging production process monitoring system includes: The high-temperature visual feature construction module is used to acquire high-speed image sequence data and mold motion state data at the hot forging station in the forging production line. Based on the high-speed image sequence data and mold motion state data, a high-temperature visual feature construction mechanism based on temperature radiation compensation is used to perform the task of extracting significant oxide scale features and output a set of candidate oxide scale feature images. The particle trajectory recognition module is used to perform the task of analyzing the motion trajectory of oxide scale particles based on the oxide scale candidate feature image set and adopts a scattering behavior recognition mechanism based on multi-scale particle trajectory evolution, and outputs a set of oxide scale scattering trajectories. The photothermal noise suppression module is used to perform abnormal particle region enhancement tasks based on the set of oxide scale scattering trajectories using a dynamic photothermal noise suppression mechanism, and outputs an oxide scale scattering region feature map. The mold cavity risk assessment module is used to perform an abnormal accumulation risk assessment task based on the abnormal accumulation discrimination mechanism based on the mold cavity risk propagation model, based on the feature map of the oxide scale scattering area, and output the abnormal accumulation risk level of oxide scale. The production monitoring module is used to make monitoring decisions for the forging process based on the risk level of abnormal oxide scale accumulation, and output the monitoring results of the forging production status.

9. A high-speed image processing and analysis-driven monitoring device for forging production processes, characterized in that, The high-speed image processing analysis-driven forging production process monitoring device includes: a memory, a processor, and a high-speed image processing analysis-driven forging production process monitoring program stored in the memory and executable on the processor. When the high-speed image processing analysis-driven forging production process monitoring program is executed by the processor, it implements a high-speed image processing analysis-driven forging production process monitoring method according to any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a high-speed image processing analysis-driven forging production process monitoring program, which, when executed by a processor, implements a high-speed image processing analysis-driven forging production process monitoring method according to any one of claims 1 to 7.

Images