Vulcanizing machine mold closing state real-time monitoring method and system based on machine vision

Abstract

The invention belongs to the technical field of image processing, and relates to a vulcanizing machine mold closing state real-time monitoring method and system based on machine vision. The method comprises the following steps: collecting a continuous image sequence in the mold closing process of a vulcanizing machine, tracking mold edge feature points and mapping the mold edge feature points into a physical height time sequence; calculating a high-order difference based on the height data, and obtaining an instantaneous speed, an acceleration and a jerk; constructing a non-linear motion retardation index containing jerk and position weight for identifying foreign matter clamping stagnation; a double-side dynamic coupling unbalance degree based on the left and right side speed difference energy ratio is constructed and is used for evaluating the mechanical balance; and comparing the index calculated in real time with a preset threshold value, and executing sudden stop protection or maintenance early warning. According to the method, through process kinematics characteristic analysis, high-sensitivity detection of tiny foreign matter retardation is realized, and accurate data support is provided for equipment maintenance.

Description

Technical Field

[0001] This invention belongs to the field of image processing technology, specifically relating to a method and system for real-time monitoring of the closed state of vulcanizing machine molds based on machine vision. Background Technology

[0002] As a core piece of equipment in the tire manufacturing process, the vulcanizing machine's mold closing motion precision directly determines the quality indicators of the finished tire, such as flash size and tread pattern misalignment, as well as the equipment's lifespan. During tire production, the mold needs to open and close frequently; any slight deviation can lead to product scrap or even equipment damage. Therefore, accurate and real-time monitoring of the vulcanizing machine's mold closing status is a crucial step in ensuring production quality and efficiency.

[0003] Currently, most mainstream mold protection technologies employ static image comparison. This method typically involves capturing an image at the moment the mold closes or is about to make contact, then comparing it with a standard template by pixel subtraction or contour comparison. By judging the difference in area or the location of feature points, the presence of foreign objects or anomalies can be determined. This method has solved the problem of obvious foreign object residue to a certain extent and has been widely used in industrial settings.

[0004] However, existing technologies have significant limitations in practical applications. First, the mold closing process of a vulcanizing machine is a dynamic process accompanied by hydraulic drive. If there is uneven lubrication of the guide rails, imbalance of thrust between the left and right hydraulic cylinders, or even the slightest soft foreign object obstruction, the mold will experience instantaneous speed fluctuations or tilting during the closing process. These dynamic characteristics will completely disappear in the static image after final closure, making such potential problems undetectable. Over time, this can lead to guide post deformation or a decrease in mold precision. Second, the vulcanizing workshop is filled with high-temperature steam, and static images are easily affected by randomly drifting steam clouds, causing local grayscale changes that are misjudged as foreign objects, resulting in a large number of false alarms. Finally, existing systems typically only output pass or fail switch results, failing to assess the health of the transmission mechanism or provide effective data support for preventative maintenance of the equipment. Summary of the Invention

[0005] The purpose of this invention is to propose a real-time monitoring method and system for the closing status of vulcanizing machine molds based on machine vision, in order to solve the technical problems of existing static photography monitoring technology being unable to identify dynamic hidden dangers during the closing process of vulcanizing machine molds, being susceptible to false alarms due to workshop steam interference, and being unable to provide predictive maintenance for equipment.

[0006] To address the above problems, the technical solution proposed in this invention for a real-time monitoring method of the closing state of a vulcanizing machine mold based on machine vision is as follows: A machine vision-based method for real-time monitoring of the closure status of vulcanizing machine molds includes the following steps: When the vulcanizing machine executes the mold closing command, it acquires a continuous image sequence of the mold closing process, uses an optical flow algorithm to track the feature points on the left and right edges of the mold, and maps the pixel coordinates of the feature points into a physical height time series with actual physical meaning. The physical height time series is subjected to high-order difference calculation to obtain the instantaneous velocity, acceleration and jerk reflecting the degree of motion change during the mold movement process; Based on the jerkiness and the real-time remaining height of the mold, a nonlinear motion retardation index is constructed. The motion retardation index uses a logarithmic function to map the jerkiness and introduces a position weighting term that increases as the real-time remaining height decreases, so as to amplify the micro-shaking characteristics at the mold closing end. A sliding time window is set, and the instantaneous speed difference between the left edge and the right edge of the mold within the sliding time window is calculated. A dual-sided coupling imbalance degree is constructed. Based on the motion lag index, it is determined whether there is foreign object obstruction to trigger emergency stop control. Based on the dual-sided coupling imbalance degree, it is determined whether there is mechanical imbalance to trigger maintenance warning, so as to realize real-time monitoring of the vulcanizing machine mold closure status based on machine vision.

[0007] Furthermore, the step of mapping the pixel coordinates of feature points to a physical height time series with actual physical meaning includes: Obtain the reference pixel coordinates of feature points in the image when the mold is fully closed; Calculate the absolute value of the difference between the vertical pixel coordinates of the feature point and the reference pixel coordinates at any given time. Multiplying the absolute value of the difference by a pre-calibrated pixel equivalent coefficient yields the vertical height of the mold from the point of complete closure at that moment.

[0008] Furthermore, the pixel equivalent coefficient is obtained as follows: During the system initialization phase, a standard-size calibration board is used to measure and calculate the actual physical length represented by one pixel.

[0009] Furthermore, obtaining the instantaneous velocity, acceleration, and jerkiness reflecting the degree of abrupt change in motion during the mold movement process includes: Calculate the height difference between the current time and the previous sampling time, and divide it by the sampling interval to obtain the instantaneous velocity; Calculate the velocity difference between the current time and the previous sampling time, and divide it by the sampling interval to obtain the acceleration; Calculate the acceleration difference between the current time and the previous sampling time, and divide it by the sampling interval to obtain the jerkiness.

[0010] Furthermore, the construction of the nonlinear motion retardation index includes: Calculate the absolute value of the jerk and perform sensitivity normalization on it; Calculate the difference between the total height of the mold's full stroke and the real-time remaining height, and use the ratio of this difference to the total height of the full stroke as the stroke percentage indicator; Based on the stroke percentage index, an index weighting term is calculated, such that the smaller the real-time remaining height of the mold, the larger the value of the index weighting term. The motion retardation index is obtained by multiplying the weighted term of the index with the normalized jerkiness, adding one, and then performing a natural logarithmic operation on the result.

[0011] Furthermore, the construction of the two-sided coupling imbalance includes: Calculate the square of the difference between the instantaneous velocity of the left edge of the mold and the instantaneous velocity of the right edge of the mold at each moment within the sliding time window, and sum them to obtain the velocity difference energy term; Calculate the sum of squares of the instantaneous velocities at the left and right edges of the mold within the sliding time window, and take the square root of their product as a normalization factor. The ratio of the velocity difference energy term to the normalization factor is calculated to obtain the bilateral coupling imbalance.

[0012] Further, the step of determining whether there is foreign object obstruction based on the motion retardation index to trigger emergency stop control includes: Determine whether the motion retardation index exceeds a set threshold for multiple consecutive frames; If so, it is determined that there is a foreign object obstructing the current mold closing process, and an emergency stop command is sent to the vulcanizing machine control system.

[0013] Furthermore, the step of determining whether mechanical imbalance exists based on the bilateral coupling imbalance degree to trigger a maintenance warning includes: Determine whether the bilateral coupling imbalance exceeds the warning threshold during the mold closing process; If so, it is determined that there is a dynamic imbalance in the left and right hydraulic drive mechanisms, and a maintenance prompt signal to check the balance of the guide column is generated.

[0014] Furthermore, the step of using an optical flow algorithm to track feature points on the left and right edges of the mold includes: Automatically lock rigid feature points on the left and right edges of the mold, the rigid feature points including the center of the positioning pin hole; The position of the rigid feature points in each frame of the image is tracked using a dense optical flow algorithm.

[0015] The technical solution of the vulcanizing machine mold closure status real-time monitoring system based on machine vision proposed in this invention is as follows: The machine vision-based real-time monitoring system for the closed state of a vulcanizing machine mold includes a processor and a memory, which stores computer program instructions. When the computer program instructions are executed by the processor, the machine vision-based real-time monitoring method for the closed state of a vulcanizing machine mold described in any of the above technical solutions is implemented.

[0016] The beneficial effects of this invention are as follows: This invention addresses the working conditions of tire vulcanization workshops filled with high-temperature steam and oil mist by employing an optical flow algorithm to track the movement of rigid feature points, rather than relying on local grayscale changes in the image. Since the flowing steam does not possess the rigid movement patterns fixed at the mold edges, the algorithm can naturally filter it out, effectively shielding it from interference from environmental factors such as steam plumes and light fluctuations, thus significantly improving the stability of the detection.

[0017] This invention, by introducing swiftness as the core monitoring element, overcomes the limitation of traditional static photography technology, which can only identify visible foreign objects. In particular, it constructs a nonlinear motion retardation index, utilizing a position weighting term that increases exponentially with decreasing real-time remaining height, giving the system the ability to dynamically adjust its sensitivity. This allows the system to automatically suppress mechanical vibration noise in the early stages of mold closure, and significantly amplify extremely small abnormal vibration signals at the final closing stage, thus ensuring that even in the presence of soft foreign objects or tiny hard particles, an emergency stop can be triggered before substantial indentation occurs.

[0018] Unlike existing technologies that only output binary results of pass or fail, the dual-sided dynamic coupling imbalance index proposed in this invention achieves real-time quantitative evaluation of the thrust balance of the left and right hydraulic cylinders and the operating status of the guide column based on the energy difference ratio algorithm. The system not only provides protection in the event of serious faults, but also detects speed asynchrony in early stages such as minor hydraulic leaks on one side or slight wear on the guide rail, issuing maintenance warnings and transforming reactive maintenance into predictive maintenance, effectively extending the equipment's service life. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the steps of the machine vision-based real-time monitoring method for the closed state of a vulcanizing machine mold according to the present invention. Figure 2 This is a schematic diagram of displacement monitoring technology in the prior art; Figure 3 This is a schematic diagram of the real-time analysis of the motion retardation index of the present invention; Figure 4 This is a schematic diagram showing the real-time comparison of the movement speeds of the two edges of the mold in an embodiment of the present invention; Figure 5 This is a schematic diagram illustrating the analysis results of the bilateral coupling imbalance in an embodiment of the present invention. Detailed Implementation

[0020] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0021] The machine vision-based real-time monitoring method for the closing state of vulcanizing machine molds proposed in this invention is mainly applied to vulcanizing machine equipment in industrial scenarios such as tire manufacturing. By deploying a high-performance image processing unit and connecting it with the industrial control system, intelligent monitoring of the mold closing process can be achieved.

[0022] like Figure 1 As shown, the real-time monitoring method for the closed state of a vulcanizing machine mold based on machine vision includes the following steps: S1. When the vulcanizing machine executes the mold closing command, it acquires a continuous image sequence of the mold closing process, uses an optical flow algorithm to track the feature points on the left and right edges of the mold, and maps the pixel coordinates of the feature points into a physical height time series with actual physical meaning.

[0023] Specifically, when the programmable logic controller (PLC) of the vulcanizing machine issues a mold-closing command, the system synchronously triggers a high-speed industrial camera mounted on the side of the vulcanizing machine to capture images. To capture minute changes during high-speed movement, the camera frame rate should be set to 100 frames per second or higher, and backlighting should be used to clearly present the mold edge contours. The system acquires a continuous image sequence of the mold closing process in real time via an image acquisition card, and automatically locks and tracks feature points on the left and right edges of the mold using a dense optical flow algorithm. These feature points are rigid feature points, preferably the centers of locating pin holes on the mold or specially marked alignment points, because rigid feature points do not shift with mold deformation. After acquiring the pixel positions of the feature points in the image coordinate system, the system needs to convert them into a physical height time series with actual physical meaning. This conversion process relies on the calibration work completed during the system initialization phase, i.e., determining the pixel equivalent coefficient using a standard-sized calibration plate. This coefficient represents the actual physical length represented by one pixel in the image, in millimeters per pixel.

[0024] The specific conversion process is as follows: First, obtain the reference pixel coordinates of the feature points in the image when the mold is in a fully closed state. Then, calculate the absolute value of the difference between the vertical pixel coordinates of the feature points at any time and the reference pixel coordinates. Finally, multiply the absolute value of the difference by the pre-calibrated pixel equivalent coefficient to obtain the vertical height of the mold from the fully closed point at that time, which is the real-time remaining height of the mold.

[0025] set up The vertical pixel coordinates of the feature point at time step are The real-time remaining height of the mold is The reference pixel coordinates are The pixel equivalent coefficient is ,but:

[0026] In the formula, It is obtained by acquiring images of the closed state during the system initialization phase.

[0027] For example, the reference pixel coordinates of the feature points are measured when the mold is fully closed during system initialization. The pixel equivalent coefficient was measured using a calibration board. mm / pixel. At a certain moment during the mold closing process. Vertical pixel coordinates of feature points in images acquired by high-speed industrial cameras Then, based on the above formula, calculate the vertical height of the mold from the point of complete closure at that moment. for: This means that the mold still has 400mm of travel left before it is fully closed. The system will continuously output two sets of time-series data, one being the height sequence of the left edge of the mold. and the height sequence of the right edge of the mold .

[0028] S2. Perform high-order difference calculation on the physical height time series to obtain the instantaneous velocity, acceleration and jerkness reflecting the degree of motion change during the mold movement process.

[0029] The system performs discretized numerical differentiation on the acquired physical height time series to obtain the instantaneous velocity, acceleration, and jerkiness reflecting the degree of motion change during the mold's movement. Specifically, this step uses the numerical difference method to calculate the height difference between the current moment and the previous sampling moment and divide it by the sampling interval to obtain the instantaneous velocity; calculate the velocity difference between the current moment and the previous sampling moment and divide it by the sampling interval to obtain the acceleration; calculate the acceleration difference between the current moment and the previous sampling moment and divide it by the sampling interval to obtain the jerkiness. This high-order difference calculation can transform minute inflection points on the displacement curve that are difficult to detect with the naked eye into significant jerkiness values, thereby keenly capturing abnormal changes in the motion process.

[0030] To more clearly and intuitively demonstrate the above calculation process, we define... The instantaneous velocity at time t is The instantaneous velocity at the previous moment was ; The acceleration at time t is The acceleration at the previous moment was ; The degree of urgency is Sampling interval is The specific calculation formula is as follows:

[0031]

[0032]

[0033] The sampling interval is determined by the frame rate of the high-speed industrial camera; for example, when the frame rate is 100 frames per second... Take 0.01s.

[0034] S3. Based on the jerkiness and the real-time remaining height of the mold, a nonlinear motion retardation index is constructed. The motion retardation index uses a logarithmic function to map the jerkiness and introduces a position weighting term that increases as the real-time remaining height decreases, so as to amplify the micro-shaking characteristics at the mold closing end.

[0035] The purpose of this step is to accurately assess the risk of foreign object obstruction by constructing a motion hysteresis index. During normal hydraulic mold closing, the mold should undergo smooth acceleration changes. If it encounters a foreign object, the smoothness of the motion trajectory will be disrupted, leading to abrupt changes in acceleration. Furthermore, the closer the mold is to closure, the lower its tolerance for foreign objects. The design logic of the motion hysteresis index lies in using a logarithmic function to compress and map the numerical range of jerkiness, and introducing a position weighting term that increases as the real-time remaining height decreases. This amplifies minute jitter characteristics at the end of mold closure, achieving segmented sensitivity control.

[0036] In this embodiment, the formula for calculating the motion resistance index is:

[0037] in, for The motion retardation index at time t, which is a dimensionless value; It is the natural logarithm function; This is the sensitivity normalization coefficient, used to adjust the weight of the effect of jerkiness on the motion retardation index; for The absolute value of the instantaneous urgency; Using natural constants An exponential function with base 0; These are weighting coefficients for hazardous areas, used to adjust the sensitivity of location weights; For step S1 Real-time remaining height of the mold (unit: mm); The total height of the mold at full stroke is denoted as , which is a fixed mechanical parameter.

[0038] It should be noted that when the mold is fully opened, The exponential term is close to , The value is set to a minimum to suppress noise interference caused by mechanical vibration during the initial mold closing stage; while when the mold is about to close... The exponential term is close to , The number of terms increases exponentially, which is intended to significantly amplify the minute abnormal vibration signals at the closed end, thereby achieving high-sensitivity protection for the mold.

[0039] To visually demonstrate the nonlinear motion retardation index proposed in this invention How to make differentiated responses to the same abnormal jitter at different closing stages is illustrated below through calculations using two sets of comparative scenarios.

[0040] Setting parameters: mm, , .

[0041] Scenario 1 is the initial stage of mold closure. Assume the mold has just begun its closing action, at which point the remaining height is relatively large, and a slight vibration occurs during the movement. Assume the remaining height at this point... mm, absolute value of jerk Substituting into the above formula, we get: .

[0042] Scenario 2 is the mold closing stage. Assume the mold is about to fully close, entering a dangerous area prone to crushing injuries from foreign objects. At this point, vibrations of the exact same amplitude as in Scenario 1 occur. Assume the remaining height at this time... mm, absolute value of jerk Substituting into the above formula, we get: .

[0043] As can be seen from the above calculations, for the same physical disturbance, the motion retardation index generated at the end of mold closure is more than 2.5 times that at the beginning stage. This significant numerical difference verifies the core advantage of the algorithm of this invention: the system can greatly improve the detection sensitivity in the critical final stage of mold closure by introducing a position weighting term that changes with the real-time remaining height, thereby ensuring that small hard objects or foreign objects can be detected in time, while effectively filtering out normal mechanical vibration interference in the start-up stage.

[0044] S4. Set a sliding time window, calculate the instantaneous speed difference between the left edge and the right edge of the mold within the sliding time window, construct the double-sided coupling imbalance degree, determine whether there is foreign object obstruction based on the motion lag index to trigger emergency stop control, and determine whether there is mechanical imbalance based on the double-sided coupling imbalance degree to trigger maintenance warning, so as to realize real-time monitoring of the vulcanizing machine mold closure status based on machine vision.

[0045] This step assesses the motion synchronization of the left and right hydraulic cylinders to prevent mold tilting. Since simple height difference comparisons are easily affected by initial installation errors, this method considers true mechanical imbalance to manifest as inconsistency in the velocity change trends of the left and right edges of the mold; therefore, the concept of energy difference ratio is introduced. Specifically, a sliding time window is set, and the degree of bilateral coupling imbalance within this window is calculated. The specific calculation formula is as follows:

[0046] In the formula, for The degree of bilateral coupling imbalance at any given moment; The sliding time window length represents the number of historical frames involved in the calculation; For time index variables; They are respectively The instantaneous velocities of the left and right edges of the mold are given by the numerator; the numerator represents the energy of the difference between the velocities of the left and right edges of the mold; in the denominator, the radical part represents the normalization factor of the total kinetic energy. To prevent odd values ​​and to prevent the denominator from being zero when the machine is stationary.

[0047] To visually demonstrate the bilateral coupling imbalance The calculation process and its physical meaning are explained below using a simplified calculation scenario. Assume a sliding time window. Set to 1, meaning only the current time is selected. and the previous moment Two frames of data were analyzed: Assume that during the monitoring process, the system collects and calculates the velocity sequence of the left edge of the mold. for Velocity sequence of the right edge of the mold for .

[0048] Calculate the numerator term reflecting the energy difference in velocity between the left and right edges of the mold, when At that time, the difference between the instantaneous velocity of the left edge and the instantaneous velocity of the right edge of the mold is The square of 1; when At that time, the difference between the instantaneous velocity of the left edge and the instantaneous velocity of the right edge of the mold is The square of is 1; therefore, the numerator is 2.

[0049] Next, calculate the denominator term representing the normalization factor for total kinetic energy. The sum of squares of the instantaneous velocities at the left edge of the mold is: The sum of the squares of the instantaneous velocities at the right edge of the mold is: The product of the two and the square root is: .

[0050] Therefore, the final bilateral coupling imbalance degree is as follows: .

[0051] This value is relatively small, indicating that the movements of the left and right edges of the mold within this sliding time window maintain a high degree of synchronization. Conversely, if the thrust of the right or left hydraulic cylinder is insufficient, resulting in a significant lag in speed, it will cause a sharp increase in the speed difference in the numerator, thereby making significantly increase.

[0052] After obtaining the above real-time indicators, the system will execute the final logic control step: input the calculated motion blockage index and the bilateral coupling imbalance degree into the logic controller. If exceeds the set threshold for N consecutive frames, the system determines that there is a foreign object blockage and immediately triggers an emergency stop protection; if exceeds the warning threshold during the mold closing process, the system determines that there is a risk of mechanical imbalance and triggers a maintenance prompt to check the balance of the guide pillars. Through this dual-index parallel monitoring mechanism, not only can the emergency stop protection for sudden failures be achieved, but also the early warning of gradual mechanical wear can be carried out, thereby effectively extending the service life of the equipment.

[0053] The following combines Figures 2-5 to further illustrate the technical solution and technical effect diagram of the present invention.

[0054] Figure 2 shows the displacement monitoring curve of the prior art. There is an extremely fine fluctuation caused by a tiny foreign object in the middle section of the curve, which is almost indistinguishable on the macroscopic displacement curve, resulting in the determination result of the prior art being qualified, thus exposing the risk of missed detection. In contrast, Figure 3 shows the motion blockage index curve calculated by using the method of the present invention. At the same time node, this curve surges instantaneously from the originally stable low position state, forming a sharp peak and significantly breaking through the safety threshold line, thereby triggering an abnormal emergency stop determination. This comparison intuitively proves that the present invention can significantly amplify the hidden tiny fault signals through the jerk and position weighting logic, effectively ensuring the detection sensitivity.

[0055] Figure 4 shows the real-time comparison of the motion speeds of the two edges of the mold. The speed curves of the left hydraulic cylinder and the right hydraulic cylinder completely overlap at the initial stage of the motion. As the process progresses, the speed curve of the right hydraulic cylinder decays and is significantly lower than the speed curve of the left hydraulic cylinder, intuitively reflecting the physical phenomenon of increased unilateral resistance or hydraulic imbalance. Figure 5The invention demonstrates the calculated bilateral coupling imbalance curve. In the initial stage of motion, the curve fluctuates steadily at a low level near zero. When a deviation occurs in the speed, the imbalance curve rapidly rises and clearly exceeds the set warning threshold. This demonstrates that the invention, through speed difference energy analysis, can characterize the left-right balance of the monitoring equipment and issue timely maintenance warnings.

[0056] Specific embodiments of the machine vision-based real-time monitoring system for the closing state of vulcanizing machine molds proposed in this invention are as follows: The machine vision-based real-time monitoring system for the closed state of a vulcanizing machine mold includes a processor and a memory. The memory stores computer program instructions. When the computer program instructions are executed by the processor, the machine vision-based real-time monitoring method for the closed state of a vulcanizing machine mold described in the above embodiments can be implemented.

[0057] The machine vision-based real-time monitoring system for the closed state of vulcanizing machine molds also includes other components well-known to those skilled in the art, such as communication buses and communication interfaces. Their settings and functions are known in the art and will not be described in detail here.

[0058] While various embodiments of the invention have been shown and described in this specification, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention.

Claims

1. A method for real-time monitoring of the mold closing status of a vulcanization press based on machine vision, characterized in that, The method comprises the following steps: When the vulcanizing machine executes the mold closing instruction, a continuous image sequence of the mold closing process is collected, feature points of the left side edge and the right side edge of the mold are tracked by using an optical flow algorithm, and pixel coordinates of the feature points are mapped into a physical height time sequence with actual physical meaning; High-order difference calculation is performed on the physical height time sequence to obtain instantaneous speed, acceleration and jerk in the mold movement process, which reflect the degree of movement mutation; Based on the jerk and the real-time residual height of the mold, a nonlinear movement resistance index is constructed, which maps the jerk by using a logarithmic function and introduces a position weighting term that increases as the real-time residual height decreases, so as to amplify the characteristics of slight shaking at the end of mold closing; A sliding time window is set, the instantaneous speed difference between the left side edge of the mold and the right side edge of the mold in the sliding time window is calculated, a bilateral coupling imbalance degree is constructed, whether there is foreign matter resistance is judged according to the movement resistance index to trigger the emergency stop control, and whether there is mechanical imbalance is judged according to the bilateral coupling imbalance degree to trigger the maintenance warning, so that real-time monitoring of the mold closing state of the vulcanizing machine based on machine vision is realized.

2. The machine vision-based mold closure condition real-time monitoring method of claim 1, wherein, The mapping of the pixel coordinates of the feature points into the physical height time sequence with actual physical meaning comprises: Obtaining the reference pixel coordinates of the feature points in the image in the completely closed state of the mold; Calculating the absolute value of the difference between the longitudinal pixel coordinates of the feature points at any time and the reference pixel coordinates; Multiplying the absolute value of the difference by a pre-calibrated pixel equivalent coefficient to obtain the vertical height of the mold from the completely closed point at this time.

3. The machine vision-based mold closure condition real-time monitoring method of claim 2, wherein, The pixel equivalent coefficient is obtained in the following manner: In the system initialization stage, a standard size calibration board is used for measurement, and the actual physical length represented by one pixel is calculated.

4. The machine vision-based mold closure condition real-time monitoring method of claim 1, wherein, The instantaneous speed, acceleration and jerk in the mold movement process are obtained in the following manner: Calculating the height difference between the current time and the last sampling time, and dividing by the sampling interval to obtain the instantaneous speed; Calculating the speed difference between the current time and the last sampling time, and dividing by the sampling interval to obtain the acceleration; Calculating the acceleration difference between the current time and the last sampling time, and dividing by the sampling interval to obtain the jerk.

5. The machine vision-based mold closure condition real-time monitoring method of claim 4, wherein, The construction of the nonlinear movement resistance index comprises: Calculating the absolute value of the jerk and performing sensitivity normalization processing thereon; Calculating the difference between the total height of the full stroke of the mold and the real-time residual height, and taking the ratio of the difference to the total height of the full stroke as a stroke proportion index; Based on the stroke proportion index, an exponential weighting term is calculated, so that the smaller the real-time residual height of the mold is, the larger the value of the exponential weighting term is; Multiplying the exponential weighting term by the normalized jerk and adding one, and then performing natural logarithm operation on the result to obtain the movement resistance index.

6. The machine vision-based real-time mold closure condition monitoring method of claim 4, wherein, The construction of the bilateral coupling imbalance degree comprises: Calculating the square of the difference between the instantaneous speed of the left side edge of the mold and the instantaneous speed of the right side edge of the mold at each time in the sliding time window, and summing to obtain a speed difference energy term; Calculating the square sum of the instantaneous speed of the left side edge of the mold and the square sum of the instantaneous speed of the right side edge of the mold in the sliding time window, and taking the square root of the product of the two as a normalization factor; The ratio of the velocity difference energy term and the normalization factor is calculated to obtain a bilateral coupling imbalance degree.

7. The machine vision-based real-time mold closure condition monitoring method of claim 5, wherein, The determination of whether there is foreign matter blockage according to the motion blockage index to trigger an emergency stop control comprises: determining whether the motion blockage index exceeds a set threshold for consecutive multiple frames; if yes, it is determined that there is foreign matter blockage in the current mold closing process, and an emergency stop command is sent to the curing press control system.

8. The machine vision-based mold closure condition real-time monitoring method of claim 6, wherein, The determination of whether there is mechanical imbalance according to the bilateral coupling imbalance degree to trigger a maintenance warning comprises: determining whether the bilateral coupling imbalance degree exceeds a warning threshold during the mold closing process; if yes, it is determined that there is dynamic imbalance in the left and right hydraulic drive mechanisms, and a maintenance prompt signal for checking the balance of the guide pillar is generated.

9. The machine vision-based real-time mold closure condition monitoring method of curing press molds as claimed in claim 1, wherein, The feature points of the left and right edges of the mold are tracked using an optical flow algorithm, comprising: automatically locking rigid feature points of the left and right edges of the mold, wherein the rigid feature points include the center of the positioning pin hole; tracking the positions of the rigid feature points in each frame of image using a dense optical flow algorithm.

10. A real-time monitoring system for mold closure status of a curing press based on machine vision, characterized in that, The method comprises a processor and a memory storing computer program instructions, which, when executed by the processor, implement the real-time monitoring method of the mold closing state of the curing press based on machine vision according to any one of claims 1-9.

Images