Thermal imaging based die casting workbench overheating warning system

Abstract

The application relates to the technical field of die-casting equipment thermal state monitoring and industrial infrared thermal imaging early warning, in particular to a die-casting workbench overheating early warning system based on thermal imaging, which comprises an infrared thermal imager and a data processing terminal; according to cooling start signals and cooling end signals, a baseline temperature matrix before cooling and a time sequence temperature field matrix after cooling are obtained, the temperature return rate of each pixel is calculated, the shielding area is recognized by combining abnormal low temperature and rapid temperature return, the real temperature field matrix is obtained by interpolation repair, and an overheating early warning result is output according to the real temperature field matrix; and the application improves the recognition ability of the shielding area of the low-radiation shielding object.

Description

Technical Field

[0001] This invention relates to the field of thermal condition monitoring of die-casting equipment and industrial infrared thermal imaging early warning technology, specifically to an overheating early warning system for die-casting workbenches based on thermal imaging. Background Technology

[0002] Currently, overheat monitoring and early warning of die-casting workbenches are usually achieved based on single-frame static temperature measurement and fixed temperature threshold determination using infrared thermal imaging. When the workbench surface is covered by a low-emissivity obstruction, the local high-temperature area in the thermal image is easily presented as an apparent low temperature. Existing methods cannot effectively identify and correct the true temperature under the obstruction area, and it is difficult to make an accurate judgment by combining the temperature change process before and after cooling, which reduces the accuracy of overheat warning and the reliability of equipment control. Summary of the Invention

[0003] The purpose of this invention is to provide an overheating early warning system for die-casting worktables based on thermal imaging, which avoids the problem of difficulty in identifying the obscured area due to the lack of dynamic reference in the same period in the prior art, and makes it easier to distinguish between the real low temperature area and the apparent low temperature area. It can accurately restore the real temperature field of the area obscured by low emissivity obscured by the obscured object to achieve a more reliable overheating early warning.

[0004] The objective of this invention can be achieved through the following technical solutions:

[0005] The overheating early warning system for the die-casting workbench based on thermal imaging includes: an infrared thermal imager for acquiring thermal radiation signals of the target area of ​​the die-casting workbench; and a data processing terminal with communication connection, including: a timing control module for outputting timing signals containing cooling start signals and cooling end signals, and receiving early warning control commands.

[0006] The thermal image acquisition module is used to continuously acquire data of the target area using an infrared thermal imager and output a raw thermal image data stream with timestamps under the control of a timing signal.

[0007] The baseline and reheat data acquisition module is used to extract the baseline temperature matrix from the original thermal map data stream based on the cooling start signal, and to obtain the time series temperature field matrix during the reheat period after cooling based on the cooling end signal.

[0008] The temperature recovery rate calculation module is used to calculate the temperature recovery rate of each pixel based on the time series temperature field matrix and output the temperature change rate matrix.

[0009] The occlusion removal module is used to jointly determine the characteristics of abnormal low temperature and rapid temperature recovery based on the baseline temperature matrix and the temperature change rate matrix, and output the occlusion binary mask matrix.

[0010] The temperature repair module is used to remove the observed temperature values ​​of the occluded area based on the occlusion binary mask matrix, and to perform interpolation repair using the temperature distribution of the normal area in the neighborhood as the boundary condition, outputting the true temperature field matrix; the overheating early warning module is used to output an early warning control command to the timing control module when it is determined that there are overheated pixels in the true temperature field matrix.

[0011] In one possible implementation, the data processing terminal also includes:

[0012] The threshold update module receives the normal area temperature recovery rate statistics from the overheat warning module, combines them with the baseline temperature matrix of the current working cycle, dynamically calculates the abnormal low temperature judgment threshold and the obstruction temperature recovery rate threshold, and sends the calculated thresholds to the obstruction removal module.

[0013] The overheat warning module is also used to feed back the statistical value of the normal area temperature recovery rate of the current working cycle to the threshold update module.

[0014] In one possible implementation, the baseline and temperature recovery data acquisition module is specifically used for:

[0015] Record the arrival time of the cooling start signal, select multiple frames of steady-state thermal images within a preset stable window before the arrival time, and generate the baseline temperature matrix corresponding to the current working cycle after time averaging.

[0016] The arrival time of the cooling end signal is recorded as the timing start point. During the cooling recovery period starting from the timing start point, multiple frames of thermal images are continuously acquired at a fixed frame rate to generate a time-series temperature field matrix corresponding to the baseline temperature matrix.

[0017] In one possible implementation, the reheat rate calculation module is specifically used for:

[0018] For each pixel coordinate in the time series temperature field matrix, obtain its temperature sequence as a function of time during the cooling and warming period;

[0019] The apparent warming rate corresponding to the pixel coordinate is obtained by linearly fitting the temperature sequence and calculating the slope.

[0020] The apparent warming rates of each pixel coordinate are combined to generate a temperature change rate matrix.

[0021] In one possible implementation, the occlusion culling module is specifically used for:

[0022] For any pixel coordinate, determine whether its temperature value in the baseline temperature matrix is ​​less than or equal to the abnormal low temperature judgment threshold to obtain the static abnormal low temperature condition.

[0023] Determine whether the temperature recovery rate of the pixel coordinate in the temperature change rate matrix is ​​greater than or equal to the temperature recovery rate threshold of the obstruction to obtain the dynamic rapid temperature recovery condition.

[0024] If and only if both static abnormal low temperature conditions and dynamic rapid temperature recovery conditions are met simultaneously, the pixel is determined to be covered by a low emissivity occluder, and the pixel is marked as an occluded area in the occlusion binary mask matrix.

[0025] If the above two conditions are not met at the same time, the pixel is determined to be a normal working platform surface, and the pixel is marked as a normal region in the occlusion binary mask matrix.

[0026] In one possible implementation, the threshold update module calculates the threshold according to the following rules:

[0027] Based on the global temperature mean and global temperature standard deviation of the current baseline temperature matrix, the abnormal low temperature judgment threshold is obtained by subtracting a deviation amount calibrated by a temperature tolerance coefficient.

[0028] The peak value of the temperature recovery slope of the normal platform material during the system's no-load operation is obtained. This peak value is then multiplied by a safety margin coefficient to obtain the threshold value of the temperature recovery rate of the obstruction.

[0029] In one possible implementation, the temperature repair module is specifically used for:

[0030] For pixels marked as normal regions in the occluded binary mask matrix, their temperature values ​​are kept as the original values ​​in the baseline temperature matrix and used as boundary conditions.

[0031] For a cell marked as an occluded area, its initial temperature value is set to the arithmetic mean of the temperatures of the surrounding normal area cells.

[0032] An iterative smoothing algorithm based on neighborhood mean is used to update the pixels in the occluded area one by one. In each update, the arithmetic mean of the temperature of the neighboring pixels around the pixel is taken as the replacement value.

[0033] The iteration stops when the temperature update magnitude of all occluded pixels in the current iteration is less than or equal to the preset convergence tolerance, and the true temperature field matrix after iteration convergence is output.

[0034] In one possible implementation, the overheat warning module is specifically used for:

[0035] Obtain the preset overheat safety critical temperature and danger judgment area threshold of the workbench material, traverse the real temperature field matrix, and find overheated pixels with temperature values ​​greater than or equal to the overheat safety critical temperature of the workbench material.

[0036] Based on the preset connected component determination rules, it is determined whether the overheated pixel forms a spatially continuous connected region, and whether the area of ​​the connected region is greater than or equal to the danger determination area threshold.

[0037] If the continuous distribution and area conditions are met, an overheating alarm result is generated, and a warning prompt, process downgrade, or shutdown inspection command is output to the timing control module.

[0038] If the continuous distribution and area conditions are not met, the current working cycle result is recorded, the statistical value of the normal area temperature recovery rate in the current working cycle is obtained and fed back to the threshold update module, and the system enters the waiting state for the next cooling start signal.

[0039] The beneficial effects of this invention are:

[0040] 1. This invention extracts the baseline temperature matrix before cooling begins and the time-series temperature field matrix during the warm-up period after cooling, calculates the warm-up rate of each pixel, and combines abnormal low temperature and rapid warm-up features to determine the shading area. This scheme overcomes the shortcomings of existing technologies that rely solely on single-frame static thermal image temperature measurement, which easily misclassifies low-emissivity shading areas as low-temperature safe areas. It fully utilizes the dynamic temperature change process before and after cooling to effectively identify shading areas.

[0041] 2. This invention, based on a binary mask matrix to remove observed temperature values ​​in shaded areas, uses the temperature distribution of the normal neighboring area as boundary conditions and employs an iterative smoothing algorithm based on the neighborhood mean for interpolation repair, outputting the true temperature field matrix. This scheme solves the problem that existing methods cannot effectively repair the true temperature under shaded areas, and can smoothly and continuously reconstruct the true heat distribution under the shading object, providing an accurate data foundation for subsequent overheating determination.

[0042] 3. This invention receives the statistical value of the normal area temperature recovery rate and dynamically calculates relevant thresholds by combining it with the baseline temperature matrix of the current working cycle. Simultaneously, based on connected component determination rules in the real temperature field matrix, it judges whether overheated pixels meet the conditions of spatial continuous distribution and area to generate an overheating alarm. This scheme overcomes the shortcomings of using fixed temperature thresholds and single-frame temperature measurement judgments, which are easily affected by changes in operating conditions. It achieves adaptive parameter adjustment across working cycles, avoids false alarms caused by single-point anomalies, and improves the accuracy of overheating early warning for the die-casting workbench and the stability of the control closed loop. Attached Figure Description

[0043] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings:

[0044] Figure 1This is a schematic diagram of a module of an overheating early warning system for a die-casting workbench based on thermal imaging, provided in an embodiment of this application. Detailed Implementation

[0045] 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.

[0046] Please see Figure 1 The overheating early warning system for a die-casting workbench based on thermal imaging includes: an infrared thermal imager for acquiring thermal radiation signals of the target area of ​​the die-casting workbench; and a data processing terminal for communication connection, including: a timing control module for outputting timing signals containing cooling start signals and cooling end signals, and receiving early warning control commands.

[0047] The thermal image acquisition module is used to continuously acquire data of the target area using an infrared thermal imager and output a raw thermal image data stream with timestamps under the control of a timing signal.

[0048] The baseline and reheat data acquisition module is used to extract the baseline temperature matrix from the original thermal map data stream based on the cooling start signal, and to obtain the time series temperature field matrix during the reheat period after cooling based on the cooling end signal.

[0049] The temperature recovery rate calculation module is used to calculate the temperature recovery rate of each pixel based on the time series temperature field matrix and output the temperature change rate matrix.

[0050] The occlusion removal module is used to jointly determine the characteristics of abnormal low temperature and rapid temperature recovery based on the baseline temperature matrix and the temperature change rate matrix, and output the occlusion binary mask matrix.

[0051] The temperature repair module is used to remove the observed temperature values ​​of the occluded area based on the occlusion binary mask matrix, and to perform interpolation repair with the temperature distribution of the normal area in the neighborhood as the boundary condition, and output the true temperature field matrix; the overheat warning module is used to output a warning control command to the timing control module when it is determined that there are overheated pixels in the true temperature field matrix.

[0052] The data processing terminal also includes: a threshold update module, which receives the normal area temperature recovery rate statistics from the overheat warning module, dynamically calculates the abnormal low temperature judgment threshold and the obstruction temperature recovery rate threshold by combining the baseline temperature matrix of the current working cycle, and sends the calculated thresholds to the obstruction removal module; the overheat warning module is also used to feed back the normal area temperature recovery rate statistics of the current working cycle to the threshold update module.

[0053] The baseline and temperature recovery data acquisition module is specifically used to: record the arrival time of the cooling start signal, select multiple frames of steady-state thermal images within a preset stable window before the arrival time, and generate the baseline temperature matrix corresponding to the current working cycle after time averaging.

[0054] The arrival time of the cooling end signal is recorded as the timing start point. During the cooling recovery period starting from the timing start point, multiple frames of thermal images are continuously acquired at a fixed frame rate to generate a time series temperature field matrix corresponding to the baseline temperature matrix.

[0055] The temperature recovery rate calculation module is specifically used to: obtain the temperature sequence of each pixel coordinate in the time series temperature field matrix as a function of time during the temperature recovery period after cooling for that pixel coordinate.

[0056] The apparent warming rate corresponding to a pixel coordinate is obtained by linearly fitting the temperature sequence to find the slope; the apparent warming rates of each pixel coordinate are combined to generate a temperature change rate matrix.

[0057] This embodiment provides an overheating early warning system for a die-casting workbench based on thermal imaging. The system uses the existing cooling process of the die-casting equipment as a unified triggering basis. First, it obtains the baseline temperature matrix before cooling, then obtains the time-series temperature field matrix after cooling, calculates the temperature recovery rate of each pixel, uses the combined features of abnormal low temperature and rapid temperature recovery to identify the occluded area, performs temperature repair on the occluded area, and outputs the overheating early warning result based on the repaired real temperature field matrix.

[0058] This scheme does not rely directly on the apparent low temperature display in a single frame thermal image, but instead combines static temperature information and dynamic temperature recovery information in the same working cycle to improve the ability to identify areas covered by low-emissivity obstructions.

[0059] In practice, the infrared thermal imager and the data processing terminal are connected through a communication interface. The infrared thermal imager continuously outputs raw thermal image data streams. After receiving the timing signal corresponding to the cooling process, the timing control module establishes a thermal image sequence index corresponding to the current work cycle in the local buffer of the data processing terminal.

[0060] The local cache is used to temporarily store the raw heat map data stream with a unified timestamp. After the heat map frames are written, they are arranged in chronological order and the corresponding cache is released after the current working cycle is completed. This ensures that the baseline temperature matrix, the time series temperature field matrix and the subsequent calculation results all come from the same working cycle and avoids cross-cycle mixing.

[0061] The original heat map data stream is acquired, and the original heat map data stream is segmented and extracted according to the cooling start signal and cooling end signal to generate a baseline temperature matrix and a time series temperature field matrix. The baseline temperature matrix refers to the two-dimensional temperature data obtained by averaging multiple frames of heat map data during the stable phase before cooling.

[0062] The time-series temperature field matrix refers to the set of multiple thermal images acquired at a fixed frame rate during the warm-up period after cooling is completed. This implementation method captures the dynamic change pattern during the temperature change process by pairing acquisitions before and after within the same working cycle, and transforms the distorted areas that are originally easily masked by emissivity differences into identifiable sources of dynamic features.

[0063] This implementation method can establish a stable temperature reference based on time-series paired acquisition, avoid misjudgment caused by accidental temperature fluctuations in a single frame, and ultimately solve the main technical defect of existing technology that makes it difficult to identify occluded areas due to the lack of dynamic references in the same period.

[0064] Obtain the cooling start signal, and according to the preset stabilization window, perform time averaging on the multiple frames of steady-state heat maps written to the buffer before the cooling start signal to generate the baseline temperature matrix corresponding to the current working cycle;

[0065] Among them, the preset stable window refers to the steady-state sampling interval established around the cooling start signal; the time average refers to averaging the temperature values ​​of the same pixel coordinates in multiple frames to reduce instantaneous jitter.

[0066] This implementation uses a multi-frame averaging method within a stable window to superimpose the same pixel, making the baseline temperature matrix better reflect the quasi-steady-state temperature distribution of the worktable surface before cooling.

[0067] This implementation method can achieve baseline temperature extraction based on stable window averaging, avoid single-frame baseline drift, and ultimately solve the main technical defect of the prior art that the subsequent discrimination is unstable due to the simple baseline establishment.

[0068] The cooling end signal is acquired, and the original heat map data stream after the cooling end signal is continuously read according to the post-cooling recovery time period and a fixed frame rate to generate a time series temperature field matrix corresponding to the baseline temperature matrix; where the post-cooling recovery time period refers to the sampling interval used to observe the thermal relaxation process after the cooling ends.

[0069] Fixed frame rate means that the infrared thermal imager outputs thermal images at a constant sampling density within this range; in this embodiment, this stage is used as an important source of information for identifying obstructions, and the differences in the warming process of each pixel are retained by continuous acquisition.

[0070] This implementation method achieves dynamic information retention through continuous sampling during the warming phase, avoiding the discarding of warming characteristics, and ultimately solves the main technical defect of the prior art that cannot distinguish between the true low temperature region and the apparent low temperature region due to ignoring the temperature change process after cooling.

[0071] Furthermore, based on the above embodiments, the temperature recovery rate calculation module can perform pixel-level slope calculation on the time series temperature field matrix;

[0072] The temperature sequence of a single pixel in the time series temperature field matrix is ​​obtained. According to the linear fitting rule, the slope of the temperature sequence of the pixel is calculated to determine the apparent warming rate of the pixel and generate a temperature change rate matrix. The temperature sequence of a single pixel refers to the set of temperature values ​​corresponding to a fixed pixel coordinate in multiple consecutive frames of heat map.

[0073] Apparent warming rate refers to the average slope of the temperature rise of a pixel over time during the warming period, and the unit can be expressed as ℃ / s. In order to quantify the difference in warming rate of each pixel after cooling, the first-order linear least squares method can be used to calculate the warming slope.

[0074] Total data collected during the cooling and warm-up period Frame heatmap, set The value range is from 1 to Positive integer sequence index, cell coordinates Apparent rate of reheat This pixel can be accessed in the first... Temperature value at the corresponding time frame With the relative timestamps of frames The calculation yields the following formula: ;

[0075] The summation symbol for each term All expressed their opinion from to Summation is performed; linear fitting is performed based on the overall trend of the entire warming process to reduce the impact of single-frame noise on slope estimation;

[0076] This embodiment uses linear fitting of the entire sequence to extract the reheat rate, so as to extract the differences in thermal response characteristics of different surfaces that have been subjected to the same cooling process.

[0077] For thin-layered shielding objects with low emissivity, the heat compensation effect below them will cause the apparent warming rate to exceed the warming rate threshold of the shielding object; for normal exposed workbench surfaces, the warming process is usually kept within the normal material warming slope reference range.

[0078] This implementation method can achieve dynamic feature extraction based on pixel-level reheat slope, avoiding the problem that differential algorithms are easily overwhelmed by noise, and ultimately solving the main technical defect of existing technologies that are difficult to identify occluded areas due to insufficient sensitivity to reheat speed.

[0079] Linear fitting is performed based on the overall trend of the entire warming process to reduce the impact of single-frame noise on slope estimation; when only the static temperature threshold is used, the low-emissivity aluminum slag covered area will directly appear as a low-temperature area and is easily classified into the low-temperature safe area.

[0080] In this embodiment, if the same pixel exhibits a low temperature during the baseline phase and a large slope during the warming phase, then the pixel will have obvious dynamic suspicious characteristics.

[0081] This implementation method transforms the post-cooling stage into an identification criterion, making it applicable to die-casting sites where there is splashing adhesion or sudden changes in emissivity.

[0082] In a preferred embodiment of the present invention, the occlusion removal module is specifically used to: for any pixel coordinate, determine whether its temperature value in the baseline temperature matrix is ​​less than or equal to the abnormal low temperature determination threshold, and obtain the static abnormal low temperature condition.

[0083] Determine whether the temperature recovery rate of the pixel coordinate in the temperature change rate matrix is ​​greater than or equal to the temperature recovery rate threshold of the obstruction to obtain the dynamic rapid temperature recovery condition.

[0084] If and only if both static abnormal low temperature conditions and dynamic rapid temperature recovery conditions are met simultaneously, the pixel is determined to be covered by a low emissivity occluder, and the pixel is marked as an occluded area in the occlusion binary mask matrix.

[0085] When the above two conditions are not met at the same time, the pixel is determined to be a normal working table surface, and the pixel is marked as a normal region in the occlusion binary mask matrix.

[0086] The threshold update module calculates the threshold according to the following rules: based on the global temperature mean and global temperature standard deviation of the current baseline temperature matrix, the abnormal low temperature judgment threshold is obtained by subtracting a deviation amount calibrated by the temperature tolerance coefficient.

[0087] The peak value of the temperature recovery slope of the normal platform material during the system's no-load operation is obtained, and the peak value of the temperature recovery slope is multiplied by the safety margin coefficient to obtain the threshold value of the temperature recovery rate of the obstruction.

[0088] The temperature restoration module is specifically used to: for pixels marked as normal regions in the occluded binary mask matrix, maintain their temperature values ​​as the original values ​​in the baseline temperature matrix, and use them as boundary conditions;

[0089] For a cell marked as an occluded area, its initial temperature value is set to the arithmetic mean of the temperatures of the surrounding normal area cells; an iterative smoothing algorithm based on the neighborhood mean is used to update the occluded cell one by one, and the arithmetic mean of the temperatures of the surrounding adjacent cells is taken as the replacement value in each update.

[0090] The iteration stops when the temperature update magnitude of all occluded pixels in the current iteration is less than or equal to the preset convergence tolerance, and the true temperature field matrix after iteration convergence is output.

[0091] Furthermore, based on the above embodiments, the occlusion removal module, the threshold update module, and the temperature repair module can work together to identify low emissivity occlusion areas from the baseline temperature matrix and the temperature change rate matrix, and to reconstruct the temperature of the area.

[0092] Obtain the baseline temperature matrix and temperature change rate matrix, and perform joint judgment on the coordinates of each pixel according to the dual threshold rule to generate an occlusion binary mask matrix.

[0093] Among them, the occlusion binary mask matrix refers to the binary labeling result that is consistent with the spatial size of the heat map. A pixel with a value of one indicates that it is suspected to be covered by a low-emissivity occluder, and a pixel with a value of zero indicates the normal working table surface.

[0094] This implementation method uses a combination of static abnormal low temperature conditions and dynamic rapid temperature recovery conditions to mark only pixels that simultaneously meet both conditions.

[0095] This implementation method can achieve occlusion recognition based on the combination of dual features, avoid false rejection caused by a single temperature threshold, and ultimately solve the main technical defect of the prior art that the occlusion area is not accurately identified due to the independent processing of static temperature.

[0096] In the specific implementation, the threshold update module can first determine two thresholds used for joint judgment;

[0097] Obtain the baseline temperature matrix of the current work cycle, and calculate the abnormal low temperature judgment threshold based on the global temperature mean and global temperature standard deviation to determine the abnormal low temperature judgment threshold.

[0098] Among them, the global temperature mean refers to the average level of temperature of all pixels within the baseline temperature matrix; the global temperature standard deviation refers to the degree of dispersion of temperature of all pixels around the mean.

[0099] By introducing a temperature tolerance coefficient To control the stringency of the abnormal low temperature judgment, the abnormal low temperature judgment threshold is set. Can be derived from the baseline temperature matrix global temperature average minus Global temperature standard deviation of the baseline temperature matrix The product is used to determine the result, and the formula is: ;

[0100] Furthermore, the peak value of the temperature recovery slope of the normal platform material during the system's no-load operation is obtained, and the temperature recovery rate threshold of the obstruction is calculated based on the safety margin coefficient to determine the temperature recovery rate threshold of the obstruction.

[0101] Among them, the peak value of the temperature recovery slope of the normal table surface material during no-load operation refers to the maximum measured value of the temperature recovery slope of the workbench material during the temperature recovery stage when it is not obstructed or contaminated.

[0102] Safety margin coefficient A preset compensation coefficient greater than 1 is used to increase the upper limit of the slope determination benchmark in order to separate and extract the shading area with abnormally rapid temperature recovery; the threshold for the temperature recovery rate of the shading object. The peak temperature rise of the normal tabletop material during no-load operation can be measured by... Multiply by the safety margin factor To determine this, the calculation formula is: ;

[0103] This implementation method adaptively adjusts two thresholds using statistics from the baseline temperature matrix and the normal material temperature recovery slope benchmark, ensuring that the judgment criteria always follow the current work cycle and equipment status changes.

[0104] This implementation method can update the threshold based on statistics and normal sample benchmarks, avoiding the insufficient sensitivity of fixed thresholds to changes in working conditions, and ultimately solving the main technical defect of poor identification stability caused by the use of undifferentiated and uniform thresholds in the existing technology.

[0105] The temperature value of a single pixel in the baseline temperature matrix and the temperature change rate matrix are obtained. Based on the abnormal low temperature judgment threshold and the occlusion temperature recovery rate threshold, the pixel is jointly judged to determine the marking result of the pixel in the occlusion binary mask matrix.

[0106] Among them, static abnormal low temperature condition refers to the baseline temperature value of the pixel being less than or equal to the abnormal low temperature judgment threshold; dynamic rapid warming condition refers to the warming rate of the pixel being greater than or equal to the warming rate threshold of the obstruction.

[0107] A pixel is marked as an occlusion area only when both conditions are met simultaneously; the static anomalous low temperature condition, which is a constraint condition that is met simultaneously, is used to characterize the apparent low temperature features formed by low emissivity occlusion.

[0108] Dynamic rapid reheating conditions are used to characterize the feature that the reheating rate of the shielding object is greater than or equal to the reheating rate threshold of the shielding object after thermal compensation by the high-temperature substrate below.

[0109] By combining static abnormal low temperature conditions and dynamic rapid temperature recovery conditions, we can eliminate the real low temperature region where the temperature recovery rate is less than the temperature recovery rate threshold of the obstruction, as well as the thermal disturbance region where the temperature value is greater than the abnormal low temperature judgment threshold.

[0110] Furthermore, after identifying the occluded area, the temperature repair module can perform interpolation repair on the occluded area.

[0111] Obtain the occlusion binary mask matrix and baseline temperature matrix, initialize the temperature field according to the different roles of normal region pixels and occluded region pixels, and generate the temperature field to be iterated.

[0112] Among them, normal region pixels refer to pixels marked as normal in the mask matrix, whose temperature values ​​are maintained as the original values ​​in the baseline temperature matrix during the restoration process and are used as boundary conditions; occluded region pixels refer to pixels marked as occluded in the mask matrix, whose initial temperature values ​​are set as the arithmetic mean of the temperatures of the surrounding normal region pixels.

[0113] This implementation method can establish the temperature field to be repaired based on the boundary constraints of the normal region, avoiding indiscriminate global smoothing of the complete thermal field, and ultimately solving the main technical defect of the prior art that the real overheating characteristics are weakened because the repair range is not distinguished.

[0114] The temperature field to be iterated is obtained, and the occluded area pixels are updated one by one according to the neighborhood mean iteration rule to generate the real temperature field matrix. The neighborhood mean iteration rule means that each time the pixel to be updated is updated, the arithmetic mean of the temperature of the neighboring pixels around it is taken as the replacement value.

[0115] The preset convergence tolerance refers to the temperature change threshold for determining whether the iteration has ended; when the temperature update amplitude of all occluded pixels in this iteration is less than or equal to the convergence tolerance, the temperature field is considered to be stable.

[0116] This process can be written in discrete iterative form: for a certain pixel within the occluded region, at the th... The temperature value in the nth iteration, determined by its temperature value in the nth iteration. The arithmetic mean of the temperatures of the surrounding neighboring pixels is obtained in the next iteration, where Let be a positive integer representing the number of iterations;

[0117] Based on the known temperature distribution in the neighboring normal region, the continuous thermal field below the shaded area is approximately restored so that the restored temperature field is consistent with the surrounding thermal distribution.

[0118] The neighborhood mean iteration process specifically adopts an eight-neighbor spatial structured decomposition: during each full-image iteration scan, the temperature value is recalculated only for the pixels in the occluded region;

[0119] For any occluded cell to be updated, the system will extract its surrounding data, including data in the horizontal, vertical, and diagonal directions. One adjacent spatial coordinate;

[0120] If a neighboring coordinate corresponds to a normal region cell, then the constant original baseline temperature value of that cell is directly used; if a neighboring coordinate corresponds to another occluded region cell, then its temperature value from the previous round is extracted. The temperature value obtained at the end of the next iteration;

[0121] This Sum the extracted temperature values ​​and divide by . The obtained value is used as the value of the occluded pixel to be updated in the 1st... The new temperature value in the next iteration;

[0122] In an iterative update, if a certain occluded pixel to be repaired is surrounded by 3 normal pixels with a stable temperature of 180℃ and 5 adjacent occluded pixels that were updated to 170℃ in the previous round, then the replacement value of the central occluded pixel after this round of update is the arithmetic mean of these 8 temperatures, which is 173.75℃.

[0123] Based on fixed boundary values ​​and adjacent pixel temperature update rules, the heat distribution in the normal area is smoothly transferred from the outer edge to the interior of the shading area.

[0124] This implementation method uses iterative updates to gradually constrain the repair results of the shading area with the surrounding heat distribution, resulting in a smoother temperature field that is closer to the original thermal topology.

[0125] This implementation method can restore the temperature of the shading area based on the neighborhood mean iteration, avoiding the sudden change in restoration results caused by simple replacement, and ultimately solving the main technical defect of the existing technology that the restoration temperature is unreliable due to the lack of continuous spatial constraints.

[0126] In a preferred embodiment of the present invention, the overheating early warning module is specifically used to: obtain a preset overheating safety critical temperature and danger judgment area threshold of the workbench material, traverse the real temperature field matrix, and find overheated pixels whose temperature value is greater than or equal to the overheating safety critical temperature of the workbench material.

[0127] Based on the preset connected component determination rules, it is determined whether the overheated pixel forms a spatially continuous connected region, and whether the area of ​​the connected region is greater than or equal to the danger determination area threshold.

[0128] If the continuous distribution and area conditions are met, an overheating alarm result is generated, and a warning prompt, process downgrade, or shutdown inspection command is output to the timing control module.

[0129] If the continuous distribution and area conditions are not met, the current working cycle result is recorded, the statistical value of the normal area temperature recovery rate in the current working cycle is obtained and fed back to the threshold update module, and the system enters the waiting state for the next cooling start signal.

[0130] Furthermore, based on the above embodiments, the overheat warning module can perform overheat determination based on the real temperature field matrix and send the determination result back to the timing control module and the threshold update module.

[0131] The real temperature field matrix is ​​obtained. Based on the preset overheating safety critical temperature of the workbench material, the real temperature field matrix is ​​traversed to find overheated pixels and generate candidate overheated regions. The overheating safety critical temperature of the workbench material refers to the upper limit temperature of safety related to the thermal fatigue risk of the workbench material.

[0132] Overheated pixels refer to pixels with a temperature value greater than or equal to the upper limit; candidate overheated areas refer to the set of overheated pixels to be determined; this implementation does not immediately issue an alarm, but continues to verify the spatial continuity and area conditions.

[0133] This implementation method can first screen candidate regions based on the over-temperature pixels in the real temperature field matrix, avoiding false alarms directly triggered by single-point temperature anomalies, and ultimately solving the main technical defect of insufficient alarm reliability caused by the coarse judgment granularity of the existing technology.

[0134] Candidate overheated regions are obtained. Based on the preset connectivity determination rules and danger determination area threshold, the spatial continuity of the candidate overheated regions is determined to determine whether an overheating alarm result is generated. The connectivity determination rules refer to the rules used to determine whether overheated pixels constitute a spatially continuous distribution area.

[0135] The connected component determination rule uses an 8-connectivity labeling algorithm to structurally decompose and identify the continuous spatial distribution: the system traverses the real temperature field matrix from left to right and from top to bottom. When it scans any candidate overheated pixel seed point that has not yet been assigned to a known connected component, it assigns it an independent region identifier, and uses this as the origin to recursively identify the surrounding regions. If adjacent pixels diverge, as long as the adjacent pixels are also candidate overheated pixels, they will be synchronized with the same identifier. The process of determining adjacent pixels continues until there are no candidate overheated pixels at the edge of the connected domain, thereby defining a spatial connected block with the same identifier.

[0136] The total number of pixels under the same identifier is accumulated and counted, and then multiplied by the pre-calibrated single-pixel real physical coverage area coefficient to obtain the actual physical area of ​​each connected domain.

[0137] The true physical coverage area coefficient of a single pixel is obtained through spatial resolution calibration using a calibration board. It represents the actual physical area of ​​space corresponding to a unit pixel, with units of 1. ;

[0138] The danger assessment area threshold refers to the minimum area required to determine that a continuous area has a risk significance; an overheating alarm result is only generated when overheated pixels form a connected area and the actual physical area of ​​the connected area exceeds the danger assessment area threshold.

[0139] This implementation method requires both continuous distribution and area conditions to be met, so that the alarm target is closer to the actual overheat concentration area.

[0140] This implementation method can achieve risk determination based on both connected domain and area conditions, avoiding false alarms caused by scattered high temperature points, and ultimately solving the main technical defect of existing technologies that cause warning distortion due to lack of spatial constraints.

[0141] Obtain overheat alarm results, and output early warning prompts, process downgrade or shutdown inspection instructions to external parties according to alarm level requirements, and determine the control results of the current work cycle;

[0142] Among them, early warning prompts can be used to remind operators to pay attention to the current thermal status of the equipment; process degradation can be used to request the upper control logic to reduce the intensity of subsequent operation; shutdown inspection commands can be used to trigger the equipment to enter the inspection state when the risk is high;

[0143] In industrial control implementation, the overheat warning module can encapsulate the above control results into corresponding control flag bits and write them into the instruction buffer read by the timing control module.

[0144] After the timing control module detects a change in the flag bit, it generates a control command corresponding to the preset control action and outputs it to the relevant control unit.

[0145] The control flag bit, once written, can enter the processed state after being read by the timing control module to prevent repeated triggering in the same work cycle; this embodiment directly converts the temperature determination result into the corresponding equipment control action.

[0146] This implementation method can realize the early warning result based on the writing of control flag bits, avoiding the alarm result remaining at the display level, and ultimately solving the main technical defect of the existing technology that causes the processing delay due to the disconnect between the result output and the device control.

[0147] Record the results of the current work cycle and obtain the statistical value of the normal area temperature recovery rate in the current work cycle. Based on the results of the current work cycle, provide feedback to the threshold update module to determine the threshold reference for the next work cycle.

[0148] Among them, the normal area temperature recovery rate statistical value refers to the temperature recovery rate statistical result corresponding to the area that is not marked as blocked and is not judged to be overheated in the current work cycle;

[0149] The overheat warning module can write the statistical value and the result of the current working cycle into the result area of ​​the local cache. The threshold update module reads and refreshes the threshold before the start of the next working cycle. After reading, the corresponding result area can be overwritten for data writing in the next working cycle.

[0150] This implementation method continuously adjusts the threshold using effective statistical values ​​from each normal working cycle;

[0151] This implementation method can update parameters based on normal sample feedback, avoiding the inadequacy of long-term fixed parameters to adapt to changes in equipment status, and ultimately solving the main technical defect of the existing technology that leads to a decrease in recognition accuracy in the later stages due to the lack of a cross-cycle adjustment mechanism.

[0152] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A die-casting worktable overheating early warning system based on thermal imaging, characterized in that, include: Infrared thermal imager is used to acquire thermal radiation signals of the target area of ​​the die-casting workbench; And a data processing terminal with communication connection, including: a timing control module, used to output timing signals containing cooling start signal and cooling end signal, and to receive early warning control commands; The thermal image acquisition module is used to continuously acquire target areas through the infrared thermal imager and output raw thermal image data streams with timestamps under the control of the timing signal; The baseline and temperature recovery data acquisition module is used to extract the baseline temperature matrix from the original heat map data stream according to the cooling start signal, and to obtain the time series temperature field matrix during the temperature recovery period after cooling according to the cooling end signal. The temperature recovery rate calculation module is used to calculate the temperature recovery rate of each pixel based on the time series temperature field matrix and output the temperature change rate matrix. The occlusion removal module is used to jointly determine the abnormal low temperature and rapid temperature recovery characteristics based on the baseline temperature matrix and the temperature change rate matrix, and output the occlusion binary mask matrix. The temperature repair module is used to remove the observed temperature values ​​of the occluded area according to the occlusion binary mask matrix, and to perform interpolation repair with the temperature distribution of the normal area in the neighborhood as the boundary condition, and output the true temperature field matrix. The overheating early warning module is used to output the early warning control command to the timing control module when it is determined that there are overheated pixels in the real temperature field matrix.

2. The overheating early warning system for die-casting worktable based on thermal imaging according to claim 1, characterized in that, The data processing terminal also includes: The threshold update module is used to receive the normal area temperature recovery rate statistics fed back by the overheat warning module, combine them with the baseline temperature matrix of the current working cycle, dynamically calculate the abnormal low temperature judgment threshold and the obstruction temperature recovery rate threshold, and send the calculated thresholds to the obstruction removal module. The overheat warning module is also used to feed back the statistical value of the normal area temperature recovery rate of the current working cycle to the threshold update module.

3. The overheating early warning system for die-casting worktable based on thermal imaging according to claim 1, characterized in that, The baseline and temperature recovery data acquisition module is specifically used for: Record the arrival time of the cooling start signal, select multiple frames of steady-state thermal images within a preset stable window before the arrival time, and generate the baseline temperature matrix corresponding to the current working cycle after time averaging. The arrival time of the cooling end signal is recorded as the timing start point. During the cooling recovery period starting from the timing start point, multiple frames of thermal images are continuously acquired at a fixed frame rate to generate a time-series temperature field matrix corresponding to the baseline temperature matrix.

4. The overheating early warning system for die-casting worktable based on thermal imaging according to claim 1, characterized in that, The temperature recovery rate calculation module is specifically used for: For each pixel coordinate in the time-series temperature field matrix, obtain its temperature sequence as a function of time during the cooling and warming-up period; The apparent warming rate corresponding to the pixel coordinate is obtained by linearly fitting the temperature sequence to find the slope. The apparent warming rates of each pixel coordinate are combined to generate the temperature change rate matrix.

5. The overheating early warning system for die-casting worktable based on thermal imaging according to claim 2, characterized in that, The occlusion removal module is specifically used for: For any pixel coordinate, determine whether its temperature value in the baseline temperature matrix is ​​less than or equal to the abnormal low temperature determination threshold to obtain the static abnormal low temperature condition. Determine whether the temperature recovery rate of the pixel coordinate in the temperature change rate matrix is ​​greater than or equal to the temperature recovery rate threshold of the obstruction to obtain the dynamic rapid temperature recovery condition. If and only if the static abnormal low temperature condition and the dynamic rapid temperature recovery condition are met simultaneously, it is determined that the pixel is covered by a low emissivity occluder, and the pixel is marked as an occluded area in the occlusion binary mask matrix. If the above two conditions are not met at the same time, the pixel is determined to be a normal working surface, and the pixel is marked as a normal region in the occlusion binary mask matrix.

6. The overheating early warning system for die-casting worktable based on thermal imaging according to claim 2, characterized in that, The specific rules for calculating the threshold by the threshold update module are as follows: Based on the global temperature mean and global temperature standard deviation of the current baseline temperature matrix, the abnormal low temperature judgment threshold is obtained by subtracting a deviation amount calibrated by a temperature tolerance coefficient. The peak value of the temperature recovery slope of the normal platform material during the system's no-load operation is obtained, and the peak value of the temperature recovery slope is multiplied by the safety margin coefficient to obtain the temperature recovery rate threshold of the obstruction.

7. The overheating early warning system for die-casting worktable based on thermal imaging according to claim 1, characterized in that, The temperature repair module is specifically used for: For pixels marked as normal regions in the binary mask matrix, their temperature values ​​are kept as the original values ​​in the baseline temperature matrix, serving as boundary conditions. For a cell marked as an occluded area, its initial temperature value is set to the arithmetic mean of the temperatures of the surrounding normal area cells. An iterative smoothing algorithm based on neighborhood mean is used to update the pixels in the occluded area one by one, and the arithmetic mean of the temperature of the neighboring pixels around the pixel is taken as the replacement value in each update. The iteration stops when the temperature update magnitude of all occluded pixels in the current iteration is less than or equal to the preset convergence tolerance, and the true temperature field matrix after iteration convergence is output.

8. The overheating early warning system for die-casting worktable based on thermal imaging according to claim 2, characterized in that, The overheat warning module is specifically used for: Obtain the preset overheating safety critical temperature and danger judgment area threshold of the workbench material, traverse the real temperature field matrix, and find overheated pixels whose temperature value is greater than or equal to the overheating safety critical temperature of the workbench material. Based on the preset connected component determination rules, it is determined whether the overheated pixel forms a spatially continuous connected region, and whether the area of ​​the connected region is greater than or equal to the danger determination area threshold. If the continuous distribution and area conditions are met, an overheating alarm result is generated, and a warning prompt, process downgrade, or shutdown inspection command is output to the timing control module. If the continuous distribution and area conditions are not met, the running result of the current working cycle is recorded, the statistical value of the normal area temperature recovery rate in the current working cycle is obtained and fed back to the threshold update module, and the system enters the waiting state for the next cooling start signal.

Images