A method for on-line determination of temperature rise residual partitioning in candy compression

Abstract

This invention relates to the field of temperature and heat measurement technology, and discloses an online method for measuring residual temperature rise in candy pressing. The method collects the start and end temperatures and coordinates of the temperature measurement positions in the center stress zone and the edge overflow zone of the mold within the candy pressing cycle measurement window, and generates a residual temperature rise sequence for each zone periodically. Based on the temperature difference fluctuation range between the two zones, the slope difference of the response curves, and the direction of periodic change, a dynamic heating rate threshold is generated, and the trigger point for spatial gradient analysis is determined. Along the adjacent order of measurement points from the center to the edge, a coherent heat transfer path is constructed based on the residual temperature rise difference and the response delay difference at the moment of maximum temperature rise, forming radial heat transfer characterization parameters. Based on the continuous decrease of these parameters, continuous heat accumulation in suspected areas, and the duration of the near-zero slope plateau, key monitoring locations are determined, and online measurement results are output. This invention can adapt to changes in thermal history during continuous pressing, characterize the direction and sequence of heat propagation, and reduce misjudgments caused by short-term fluctuations.

Description

Technical Field

[0001] This invention relates to the field of temperature and heat measurement technology, specifically to an online method for determining the residual temperature rise zone in candy pressing. Background Technology

[0002] Candy pressing production typically involves repeated pressing cycles. The central stress zone of the mold bears the main pressing load, while the edge overflow zone contacts the edge of the candy and the mold wall. These two zones differ in terms of stress, friction, and heat dissipation boundaries. As the pressing cycle continues, the localized temperature rise generated in the current cycle is superimposed on subsequent cycles, causing the mold temperature to dynamically change with running time and spatial location.

[0003] Existing technologies typically involve placing temperature sensing elements at different locations within the mold to collect temperature time series data at each location. Filtering, temperature interpolation, heat conduction models, or fixed threshold alarms are then used to obtain the temperature field distribution or single-point exceeding-limit status. Some solutions also determine whether local temperature anomalies are based on the temperature difference between adjacent measuring points or the temperature rise change within a single time window.

[0004] However, under repeated candy pressing conditions, the background temperature and thermal history will change with the cycle, and a fixed threshold is difficult to distinguish between normal heat accumulation and continuous anomalies; the scalar temperature distribution formed by sparse measuring points is difficult to reflect the direction and sequence of heat propagation from the central stress area to the edge overflow area; relying solely on the parameter changes within a single temperature difference or window, it is easy to misjudge sampling noise, short-term operating condition fluctuations, or temporary plateaus as continuous anomalies, making it difficult to form verifiable measurement results for key monitoring locations.

[0005] The information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0006] The present invention aims to overcome the above-mentioned problems and provides an online method for measuring residual temperature rise in candy pressing, so as to solve the defects in the prior art that it is difficult to accurately reflect the direction of heat propagation, propagation time sequence and continuous abnormal state under continuous pressing conditions.

[0007] The technical solution of this invention is as follows:

[0008] A method for online determination of residual temperature rise in candy pressing zones includes: within a candy pressing cycle measurement window, collecting the start and end temperatures and coordinates of temperature measurement positions in the center stress zone and the edge overflow zone of the mold; calculating the temperature rise at the measurement positions and the temperature rise difference between the two zones; accumulating the temperature rise periodically to generate a residual temperature rise sequence; generating a dynamic heating rate threshold based on the temperature difference fluctuation range of the two zones during the baseline sampling stage, the slope difference of the response curve, and the direction of periodic change; marking the temperature measurement positions in the center stress zone where the ratio of the accumulated residual temperature rise to the measurement window duration continuously exceeds the threshold as spatial gradient analysis trigger points; calculating the residual temperature rise difference between adjacent measurement points and the response delay difference between the maximum temperature rise time according to the adjacent order of measurement points from the center stress zone to the edge overflow zone; and classifying the two types of residual temperature rise into a spatial gradient analysis sequence. A continuous heat transfer path is obtained by connecting measurement points with positive differences and continuous adjacent order and temperature measurement sequence. Radial heat transfer characterization parameters are formed based on the path direction of the continuous heat transfer path, the radial direction from the spatial gradient analysis trigger point to the temperature measurement position of the edge overflow area, and the distance between measurement points. The path segment where the parameter continuously decreases and the cumulative decrease exceeds the decrease threshold in multiple consecutive sliding windows is identified as a suspected conduction blockage area. When the cumulative residual temperature rise of the suspected conduction blockage area is higher than that of the adjacent normal area in multiple consecutive pressing cycles, and the slope of its response curve is continuously in the preset near-zero interval for a duration exceeding the platform duration threshold, the corresponding position is identified as a key monitoring position, and the online measurement result of heat conduction blockage is output.

[0009] Optionally, generating the residual temperature rise sequence for each zone includes: calculating the difference between the end temperature and the start temperature of each temperature measurement position in the central stress zone and the edge overflow zone within the measurement window to obtain the temperature rise at the measurement position; calculating the temperature rise difference between corresponding temperature measurement positions in the two zones to obtain the zone temperature difference value; accumulating the temperature rise at the same temperature measurement position according to the pressing cycle sequence to obtain the cumulative residual temperature rise, and associating the cumulative residual temperature rise, the zone temperature difference value, and the coordinates according to the pressing cycle sequence to obtain the residual temperature rise sequence for each zone.

[0010] Optionally, before generating the dynamic heating rate threshold, the method further includes correcting the residual temperature rise sequence of the partition. The correction includes: using Kalman filtering to process the continuous temperature values ​​at the same temperature measurement location and the synchronous temperature values ​​of adjacent measurement points to obtain smoothed temperature values ​​and smoothed synchronous values; calculating the reference deviation between the smoothed temperature value and the smoothed synchronous value; when the reference deviation exceeds the deviation threshold, using the smoothed synchronous value to correct the smoothed temperature value at the corresponding temperature measurement location to obtain the target temperature value; recalculating the temperature rise and residual temperature rise accumulation at the corresponding temperature measurement location based on the target temperature value; establishing a spatial topology map based on the coordinates of each temperature measurement location and the adjacency relationship of the measurement points; and using Kriging interpolation to perform spatial gridding compensation on the residual temperature rise accumulation.

[0011] Optionally, the process of obtaining the temperature difference fluctuation range between the two zones includes: calculating the temperature rise difference between the central stress zone and the edge overflow zone to form a benchmark sampling temperature difference sequence; removing outliers from the benchmark sampling temperature difference sequence and calculating the upper and lower limits of the remaining temperature difference values; and forming a temperature rise rate threshold calculation benchmark corresponding to the benchmark sampling stage based on the upper and lower limits.

[0012] Optionally, generating the dynamic heating rate threshold includes: fitting the temperature response curves of the central stress zone and the edge overflow zone within the current pressing cycle to obtain a first temperature rise slope and a second temperature rise slope; calculating the difference between the first temperature rise slope and the second temperature rise slope, and performing differential analysis on the temperature difference sequence of the continuous pressing cycle to obtain the direction of cycle change; classifying the combination state according to whether the slope difference is greater than zero, the direction of cycle change, and the position of the current partition temperature difference value relative to the upper and lower limits of the temperature difference fluctuation range of the two partitions; determining the 95th percentile of the residual temperature rise rate in the historical effective pressing cycle that is the same as the current combination state as the dynamic heating rate threshold corresponding to the current pressing cycle; when there are insufficient historical effective samples of the corresponding combination state, the previous effective dynamic heating rate threshold is used.

[0013] Optionally, marking the spatial gradient analysis trigger point includes: calculating the ratio of the cumulative residual temperature rise at each temperature measurement location to the measurement window duration to obtain the residual temperature rise rate; comparing the residual temperature rise rate with the dynamic heating rate threshold point by point, and marking the temperature measurement locations in the central stress area that continuously exceed the dynamic heating rate threshold as spatial gradient analysis trigger points; determining, according to the adjacent order of the measurement points, target temperature measurement locations in the edge overflow area that are reachable from the spatial gradient analysis trigger point along the adjacent order of the measurement points and have a clear adjacent order, and calculating the coordinate difference, temperature rise difference, and response delay between the two locations when the maximum temperature rise is reached.

[0014] Optionally, obtaining a coherent heat transfer path includes: calculating the residual temperature rise difference and response delay difference between adjacent measuring points according to the adjacent order of measuring points from the central stress area to the edge overflow area; removing two types of differences whose absolute values ​​do not exceed their respective noise tolerances; determining whether the two types of differences are both positive along the adjacent order of the measuring points, and connecting adjacent measuring points that are both positive and have continuous temperature measurement sequences as candidate node strings; when there are multiple candidate node strings, sorting them in descending order according to the number of consecutive effective connections, the minimum residual temperature rise difference within the node string, and the minimum response delay difference, and selecting the candidate node string with the highest sorting result; if the sorting results are still the same, selecting the candidate node string whose starting measuring point is closest to the spatial gradient analysis trigger point to obtain the coherent heat transfer path.

[0015] Optionally, the formation of radial heat transfer characterization parameters includes: calculating the ratio of the temperature change rate of the spatial gradient analysis trigger point to the temperature measurement position of the corresponding edge overflow area within the measurement window to obtain an initial rate ratio; obtaining the path direction vector of the continuous heat transfer path and the radial direction vector from the spatial gradient analysis trigger point to the temperature measurement position of the edge overflow area, and determining the directional projection coefficient based on the two direction vectors; using the product of the initial rate ratio and the directional projection coefficient as the radial correction ratio, and performing distance correction on the radial correction ratio based on the distance between the measurement points of the two temperature measurement positions to obtain the radial heat transfer characterization parameters.

[0016] Optionally, determining the suspected conduction blockage region includes: calculating the change in radial heat transfer characterization parameters of each path segment of the continuous heat transfer path using a sliding window; when the radial heat transfer characterization parameters of the same path segment decrease in multiple consecutive sliding windows, and the cumulative decrease exceeds the decrease threshold determined according to the baseline sampling stage, the path segment is marked as an abnormal segment; the coordinates, zone temperature difference values, and residual temperature rise accumulation of the abnormal segment are associated, and a spatial feature matrix is ​​constructed after normalizing the three types of data; the spatial feature matrix is ​​then processed using a mean-based clustering algorithm to obtain the spatial distribution of the suspected conduction blockage region.

[0017] Optionally, determining the corresponding location as a key monitoring location includes: calculating the difference in residual temperature rise between the suspected conduction blockage area and the adjacent normal area within the same suppression cycle; generating a continuous heat accumulation indicator when the difference exceeds the accumulation threshold in multiple consecutive suppression cycles; smoothing the temperature response curve of the suspected conduction blockage area and calculating the slope sequence; determining that the temperature response curve has a plateau segment when the slope sequence is continuously in a preset near-zero interval for a duration exceeding a plateau duration threshold, and determining the location corresponding to the plateau segment as a key monitoring location.

[0018] The beneficial effects of this invention are as follows:

[0019] This invention collects the start and end temperatures and coordinates of the temperature measurement positions in the central stress zone and the edge overflow zone within the measurement window of the candy pressing cycle, and accumulates them periodically to form a residual temperature rise sequence for each zone, so that the measurement results reflect the thermal history changes during continuous pressing. By combining the temperature difference fluctuation range of the two zones, the slope difference of the response curve, and the direction of periodic change, a dynamic heating rate threshold is generated, and the trigger point for spatial gradient analysis is determined by the continuous exceeding of the residual temperature rise rate, reducing the influence of background temperature changes on the fixed threshold. A coherent heat transfer path is constructed by the temperature rise residual difference and the response delay difference at the maximum temperature rise moment, and radial heat transfer characterization parameters are formed by the path direction, radial direction, and measurement point spacing, so that the discrete measurement point data reflects the direction and sequence of heat propagation. Finally, by continuously decreasing the window, continuously cyclically accumulating heat, and jointly confirming the duration of the near-zero slope plateau, misjudgments caused by short-term fluctuations and temporary plateaus are reduced, forming verifiable measurement results for key monitoring locations. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall processing flow for online determination of residual temperature rise during candy pressing according to an embodiment of the present invention;

[0021] Figure 2 This is a schematic diagram of the temperature measurement positions of the mold partitions and the adjacency relationship of the measurement points from the center to the edge, provided in one embodiment of the present invention;

[0022] Figure 3 This is a schematic diagram of the formation of residual temperature rise sequence, data correction and spatial gridding compensation provided in one embodiment of the present invention;

[0023] Figure 4 This is a schematic diagram illustrating the dynamic heating rate threshold generation and spatial gradient analysis trigger point determination provided in one embodiment of the present invention;

[0024] Figure 5 This is a schematic diagram of the construction of a coherent heat transfer path and the formation of radial heat transfer characterization parameters provided in one embodiment of the present invention;

[0025] Figure 6 This is a schematic diagram illustrating the joint determination of suspected conduction blockage areas and key monitoring locations provided in one embodiment of the present invention;

[0026] Figure 7 This is a schematic diagram of the physical arrangement of the temperature measurement positions of the mold and the adjacent path from the center to the edge provided in one embodiment of the present invention;

[0027] Figure 8 This is a schematic diagram of the spatial gridded distribution of residual temperature rise accumulation provided in one embodiment of the present invention;

[0028] Figure 9This is a schematic diagram of residual temperature rise accumulation interpolation grid node cloud map and coherent heat transfer path provided in one embodiment of the present invention;

[0029] Figure 10 This is a partially enlarged schematic diagram of the spatial distribution of residual temperature rise in the vicinity of a key monitoring location provided in an embodiment of the present invention. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining the invention and are not intended to limit the invention; that is, the described embodiments are merely some embodiments of the invention, and not all embodiments. The components of the embodiments of the invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0031] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0032] It should be noted that relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0033] As mentioned earlier, during continuous candy pressing, the mold background temperature and thermal history change with the pressing cycle. Existing fixed temperature thresholds are insufficient to distinguish between normal heat accumulation and persistent anomalies requiring further analysis. At the same time, limited and discrete temperature measurement locations can usually only reflect the local temperature or temperature difference at each measurement location, making it difficult to explain the direction and sequence of heat propagation from the central stress area to the edge overflow area. If judgment is made based solely on a single temperature difference or parameter changes within a single window, sampling noise, short-term operating condition fluctuations, or temporary plateaus are easily misjudged as persistent anomalies, making it difficult to form verifiable measurement results for key monitoring locations.

[0034] To address this, the present invention provides an online method for measuring residual temperature rise in candy pressing. The method places the periodic temperature rise, cumulative residual temperature rise, maximum temperature rise time, location coordinates, and adjacency relationship of the measuring points in the same processing chain, thereby forming a dynamic heating rate threshold, spatial gradient analysis trigger point, continuous heat transfer path, radial heat transfer characterization parameters, suspected conduction blockage area, and key monitoring location.

[0035] The following is combined with Figures 1 to 6 This invention is described in detail.

[0036] Example 1:

[0037] like Figure 1 As shown, the online measurement method for residual temperature rise in candy pressing provided by this embodiment of the invention is executed in the order of steps S100 to S500, sequentially completing the generation of residual temperature rise sequence in each zone, determination of dynamic heating rate threshold and spatial gradient analysis trigger point, construction of coherent heat transfer path, determination of radial heat transfer characterization parameters and suspected conduction hindrance region, determination of key monitoring locations, and output of online measurement results; the executing entity can be a synchronous temperature acquisition device connected to each temperature measurement location of the mold and an online measurement processor. The method mainly includes the following steps:

[0038] S100. Within the candy pressing cycle measurement window, collect the start and end temperatures and coordinates of the temperature measurement positions in the center stress area and the edge overflow area of ​​the mold, calculate the temperature rise at the temperature measurement position and the temperature rise difference between the two zones, and accumulate the temperature rise according to the cycle to generate a zone temperature rise residual sequence.

[0039] S200: Based on the temperature difference fluctuation range of the two zones in the benchmark sampling stage, the slope difference of the response curve, and the direction of periodic change, a dynamic heating rate threshold is generated. The temperature measurement position of the central stress zone that continuously exceeds the ratio of the residual temperature rise to the measurement window duration is marked as the spatial gradient analysis trigger point.

[0040] S300. According to the adjacent order of the measuring points from the central stress area to the edge overflow area, calculate the residual temperature rise difference between adjacent measuring points and the response delay difference between the maximum temperature rise time. Connect the measuring points with both types of differences being positive and continuous along the adjacent order of the measuring points and with continuous temperature measurement time to obtain a coherent heat transfer path.

[0041] S400. Based on the path direction of the continuous heat transfer path, the radial direction from the spatial gradient analysis trigger point to the temperature measurement position of the edge overflow area, and the distance between the measurement points, a radial heat transfer characterization parameter is formed. The path segment in which the parameter continuously decreases within multiple consecutive sliding windows and the cumulative decrease exceeds the decrease threshold is identified as a suspected conduction blockage area.

[0042] S500 When the cumulative residual temperature rise in the suspected conduction blockage area is higher than that in the adjacent normal area for multiple consecutive suppression cycles, and the slope of its response curve remains in the preset near-zero range for a duration exceeding the platform duration threshold, the corresponding location is identified as a key monitoring location, and the online measurement results of heat conduction blockage are output.

[0043] Based on the above steps, this invention sequentially correlates the residual temperature rise of periodic zones, dynamic rate triggering, directional judgment of double differences between adjacent measuring points, continuous decrease of radial heat transfer characterization parameters, and continuous heat accumulation with the duration of the platform, so that the temperature data of discrete measuring points can form online measurement results with spatial direction, propagation sequence and duration conditions, thereby reducing the impact of fixed thresholds, single differences or single window fluctuations on the process of determining key monitoring locations.

[0044] Example 2:

[0045] To provide a more detailed explanation of the technical solutions provided in the above embodiments, the present invention also provides another preferred embodiment. A synchronous temperature acquisition device collects temperature data under a unified time reference and transmits the temperature record, which includes measurement point identification, area identification, location coordinates, compression cycle number, sampling timestamp, and valid status, to an online measurement processor. The online measurement processor processes the data in the following order: residual temperature rise formation, data correction, threshold generation, trigger point determination, path construction, radial characterization, persistence judgment, and result output.

[0046] In this embodiment, as Figure 7 As shown, the measurement window duration is determined by the compression cycle trigger flag and the on-site initialization configuration (e.g., the measurement window duration is 4.0s), and the sampling interval is determined by the sampling capability of the synchronous temperature acquisition unit and the response time of the temperature measurement channel (e.g., the sampling interval is 0.1s). Each measurement window is located within a single compression cycle. After the window start sample, window end sample, cycle number, and timestamp satisfy the corresponding relationship, the temperature data enters the subsequent calculation.

[0047] like Figure 2 As shown, the mold temperature measurement points are arranged in zones according to the central stress zone and the edge overflow zone. Measurement point C1 in the central stress zone, intermediate measurement points M1 and M2, and measurement point E1 in the edge overflow zone are arranged according to... The directions form an example adjacency chain that runs through this embodiment. The illustrated partition boundary, coordinate direction, adjacent direction of measurement points, and distance between measurement points serve as the spatial basis for subsequent calculations of temperature rise difference, response delay, path direction, and distance between measurement points.

[0048] like Figure 3 As shown, in step S100, the formation of the partitioned temperature rise residual sequence may further include:

[0049] S110. Calculate the difference between the end temperature and the start temperature of each temperature measurement position in the central stress zone and the edge overflow zone within the measurement window to obtain the temperature rise at the measurement position.

[0050] S120. Calculate the temperature rise difference between the corresponding temperature measurement positions in the two zones to obtain the temperature difference value between the zones.

[0051] S130. Accumulate the temperature rise of the same temperature measurement position according to the pressing cycle to obtain the residual temperature rise accumulation. Then, associate the residual temperature rise accumulation, the temperature difference value of the partition, and the coordinates according to the pressing cycle to obtain the partition temperature rise residual sequence.

[0052] For example, steps S110 to S130 may further include the following processes:

[0053] First, the synchronous temperature acquisition unit records the first valid temperature at each measurement location at the start of the measurement window and the last valid temperature before the end of the measurement window. The online measurement processor reads the start and end temperatures according to the same measurement point, the same cycle, and the same time reference, and calculates the temperature rise at the measurement location using the unified direction of subtracting the start temperature from the end temperature. If the start or end temperature is missing, the timestamp is in reverse order, or the temperature value exceeds the range of the temperature measurement channel, the corresponding measurement point is marked as invalid in the current cycle and is not replaced with a zero value.

[0054] Secondly, the temperature measurement positions in the central stress zone and the edge overflow zone are paired according to the coordinate correspondence written during initialization. The pairing relationship remains unchanged within the same operating phase, and no other measurement points are temporarily used if a corresponding point is missing. The online measurement processor calculates the difference in temperature rise between the corresponding measurement positions of the two zones to obtain the zone temperature difference value, and saves this difference value along with the cycle number and the coordinates of the two measurement points.

[0055] Then, the online measurement processor updates the cumulative residual temperature rise at the same temperature measurement location according to the pressing cycle number. During mold changing, recalibration, or re-establishment of the reference stage, the cumulative residual temperature rise can be set to zero and the reset time recorded; invalid cycles only retain the notch mark and are not included in the accumulation. The residual temperature rise rate is expressed by equation (1):

[0056] (1)

[0057] In equation (1), Number the suppression cycle; Number the temperature measurement locations; and Temperature measurement locations In the The starting and ending temperatures within the measurement window for each pressing cycle are measured in °C. This represents the cumulative residual temperature rise at the temperature measurement location after the end of the previous effective pressing cycle, expressed in °C. The duration of the measurement window is measured in seconds (s). This represents the residual temperature rise rate for the current cycle, expressed in °C / s.

[0058] The numerator of equation (1) is the cumulative residual temperature rise after the current cycle update. This cumulative amount is used for subsequent calculations of residual temperature rise difference, spatial characteristics, and continuous heat accumulation. The residual temperature rise rate is used for comparison with the dynamic heating rate threshold. The residual temperature rise rate referred to in this specification is the cumulative temperature rise intensity index obtained by dividing the cumulative residual temperature rise formed according to the same cumulative rule by the current measurement window duration. It does not only represent the instantaneous heating slope of the temperature response curve of the current pressing cycle. The dynamic heating rate threshold is determined from historical effective pressing cycles using the same cumulative caliber as the residual temperature rise rate.

[0059] Continuing with the previous example, the starting temperature of C1 in the current cycle is 31.0℃ and the ending temperature is 32.5℃, while the starting temperature of E1 is 31.0℃ and the ending temperature is 31.7℃. Therefore, the temperature rise at the measurement locations of C1 and E1 is 1.5℃ and 0.7℃, respectively, with a zone temperature difference of 0.8℃. If the cumulative residual temperature rise of C1 before the current cycle is 7.7℃, then the cumulative residual temperature rise after the update in the current cycle is 9.2℃. Substituting this value into the 4.0s measurement window duration, the residual temperature rise rate is 2.30℃ / s. The processor associates 9.2℃, 0.8℃, the coordinates of C1, and the current cycle number to form a valid record in the zone temperature rise residual sequence.

[0060] In this second embodiment, after forming the partitioned temperature rise residual sequence in step S130 and before generating the dynamic temperature rise rate threshold in step S200, the following correction process can be further performed:

[0061] S140. Kalman filtering is used to process the continuous temperature values ​​at the same temperature measurement location and the synchronous temperature values ​​at adjacent measurement points to obtain smoothed temperature values ​​and smoothed synchronous values.

[0062] S150. Calculate the reference deviation between the smoothed temperature value and the smoothed synchronization value. When the reference deviation exceeds the deviation threshold, use the smoothed synchronization value to correct the smoothed temperature value at the corresponding temperature measurement position to obtain the target temperature value.

[0063] S160. Recalculate the temperature rise and residual temperature rise accumulation at the corresponding temperature measurement location based on the target temperature value. Establish a spatial topology map based on the coordinates of each temperature measurement location and the adjacency relationship of the measurement points. Use Kriging interpolation to perform spatial gridding compensation on the residual temperature rise accumulation.

[0064] For example, steps S140 to S160 may further include the following processes:

[0065] First, the calibration processor reads continuous temperature values ​​from the same temperature measurement location and synchronous temperature values ​​from adjacent measurement points at the same sampling time, and performs one-dimensional Kalman filtering with temperature as the observed quantity. The process noise variance is taken as the variance of adjacent effective temperature increments at the same temperature measurement location during the baseline phase, and the observation noise variance is taken as the residual variance of the measured values ​​relative to the calibration temperature under the calibrated isothermal state (e.g., process noise variance is 0.0025℃², and observation noise variance is 0.0100℃²). Both types of temperature sequences are filtered separately to avoid eliminating the original temperature differences between adjacent measurement points before deviation comparison.

[0066] Secondly, the processor acquires the synchronous temperature difference benchmark value between the same temperature measurement location and the corresponding adjacent reference measurement point in the baseline stage, and calculates the absolute deviation of the difference between the current smoothed temperature value and the smoothed synchronous value relative to the synchronous temperature difference benchmark value. This absolute deviation is used as the benchmark deviation and compared with a deviation threshold. The deviation threshold is jointly determined by the allowable error of the temperature measurement channel calibration and the synchronous differential distribution in the baseline stage (for example, the deviation threshold is 0.30℃). When the benchmark deviation exceeds the deviation threshold and the adjacent reference measurement point is valid, the smoothed temperature value at this point is corrected using the smoothed synchronous value; when the benchmark deviation does not exceed the deviation threshold, the smoothed temperature value at this point is retained; when the adjacent reference measurement point is also invalid, no alternative correction is implemented.

[0067] Then, as Figure 8 As shown, the processor recalculates the temperature rise and residual cumulative temperature rise at the corresponding temperature measurement location based on the target temperature value, and establishes a spatial topology map based on the measurement point coordinates and the fixed adjacency relationship from the center to the edge. Nodes in the spatial topology map store measurement point identifiers, region identifiers, coordinates, and valid states; edges between nodes store adjacency directions and measurement point spacing. The grid spacing for Kriging interpolation is determined by the mold measurement point design spacing (e.g., a grid spacing of 2 mm). The lower limit of the number of valid measurement points participating in the spatial variogram fitting is determined by the leave-one-out verification results from the baseline stage, i.e., the minimum number of valid measurement points when the error between the interpolation result and the measured value of the left-out measurement point does not exceed the allowable error of the temperature measurement channel calibration (e.g., a lower limit of 6 valid measurement points). Interpolation only performs spatial gridding compensation for the residual cumulative temperature rise within the area enclosed by the valid measurement points, without extrapolating and filling areas without adjacency basis.

[0068] Following the above Regarding the measurement point layout, here's another example of calibration timing within a window: At the same sampling moment, the smoothed temperature value of C1 is 32.1℃, and the smoothed synchronous value of the adjacent reference measurement point M1 is 31.7℃. The current synchronous temperature difference relative to the reference value of the synchronous temperature difference at this measurement point is 0.4℃, which is greater than the deviation threshold of 0.30℃. Since M1 is in a valid state, 31.7℃ is taken as the target temperature value for C1. Based on this, the processor recalculates the temperature rise and residual temperature rise accumulation of C1 in the current cycle, and establishes C1, M1, M2, and E1 sequentially as... The adjacent chain. If the number of effective measuring points meets the interpolation requirements, a compensated residual temperature rise distribution is formed on a 2mm grid; if the number of effective measuring points is insufficient, the results of discrete measuring points are retained and the uncovered areas are marked. The example of the correction time in this section is used to illustrate the synchronous correction rule. The subsequent calculations involving 9.2℃, 2.30℃ / s and the cumulative residual temperature rise of adjacent measuring points will continue to use the aforementioned periodic through-the-loop example and will not be combined with the calculation of the correction time example in this section.

[0069] Next, when generating the dynamic heating rate threshold in step S200, the temperature difference fluctuation range between the two zones can be obtained through the following process:

[0070] S210, calculate the temperature rise difference between the central stress zone and the edge overflow zone to form a benchmark sampling temperature difference sequence;

[0071] S220. Remove outliers from the benchmark sampled temperature difference sequence and calculate the upper and lower limits of the remaining temperature difference values;

[0072] S230. Based on the upper and lower limits, a temperature rise rate threshold calculation benchmark is formed corresponding to the benchmark sampling stage.

[0073] For example, steps S210 to S230 may further include the following processes:

[0074] First, the baseline sampling phase begins after the mold has completed initialization and the measurement window is continuously valid. The number of effective pressing cycles included in the baseline sampling phase is determined by the ratio of the allowable initialization time given by the on-site initialization configuration to the duration of a single pressing cycle. After eliminating invalid cycles, sampling continues until the corresponding number is reached (for example, if the allowable initialization time is 80s and the duration of a single pressing cycle is 4s, the baseline sampling phase includes 20 effective pressing cycles). The online measurement processor uses a fixed partition correspondence to calculate the temperature rise difference between the central stress zone and the edge overflow zone cycle by cycle, and forms a baseline sampling temperature difference sequence according to the cycle number.

[0075] Secondly, the processor uses an isolated forest algorithm to screen outliers in the baseline sampled temperature difference sequence. The outlier score threshold is determined by the high quantile of the baseline valid samples (e.g., the outlier score threshold is the 95th percentile of the baseline outlier score, corresponding to an example value of 0.62). Temperature difference values ​​exceeding this threshold are not included in the upper and lower limit statistics. The processor uses the maximum and minimum values ​​of the remaining valid temperature difference values ​​as the upper and lower limits of temperature difference fluctuation, respectively. If the valid samples are insufficient after removing outliers, the baseline sampling phase is extended without updating the temperature difference fluctuation range.

[0076] Then, the processor retains the original values ​​and units of the upper and lower limits of temperature fluctuation and performs scaling according to the benchmark range to form the benchmark for calculating the temperature rise rate threshold corresponding to the current mold. Scale scaling is only used for subsequent segmented mapping rule input and does not change the original temperature physical quantity.

[0077] Using the previous example, after anomaly screening, the temperature difference values ​​of 20 effective pressing cycles have an upper limit of 0.45℃, a lower limit of 0.10℃, and a temperature difference range of 0.35℃. The temperature difference of the current cycle zone is 0.8℃, which is higher than the upper limit of the baseline temperature difference fluctuation. Based on this, the processor enters the dynamic heating rate threshold update process.

[0078] like Figure 4 As shown, in this second embodiment, after obtaining the calculation benchmark for the temperature rise rate threshold, the dynamic temperature rise rate threshold in step S200 can be further generated through the following process:

[0079] S240. Fit the temperature response curves of the central stress area and the edge overflow area during the current pressing cycle to obtain the first temperature rise slope and the second temperature rise slope.

[0080] S250. Calculate the difference between the first temperature rise slope and the second temperature rise slope, and perform differential analysis on the temperature difference sequence of the continuous pressing cycle to obtain the direction of periodic change.

[0081] S260. Based on whether the slope difference is greater than zero, the direction of the period change, and the position of the current partition temperature difference value relative to the upper and lower limits of the temperature difference fluctuation range of the two partitions, the combination state is divided. The 95th percentile of the residual temperature rise rate in the historical effective suppression cycle that is the same as the current combination state is determined as the dynamic temperature rise rate threshold corresponding to the current suppression cycle. When the historical effective samples of the corresponding combination state are insufficient, the previous effective dynamic temperature rise rate threshold is used.

[0082] For example, steps S240 to S260 may further include the following processes:

[0083] First, the online measurement processor reads the effective temperature-time samples of the central stress zone and the edge overflow zone within the current pressing cycle, and fits the temperature response curves of the two zones using the least squares method without crossing the measurement window boundaries. If the number of sampling points is insufficient or the timestamps are discontinuous, the slope for the current cycle is not generated.

[0084] Secondly, the processor obtains the slope difference by subtracting the second temperature rise slope from the first temperature rise slope, and obtains the cycle difference by subtracting the temperature difference of the previous valid cycle from the temperature difference of the current cycle zone. When the cycle difference is positive, zero, or negative, it forms the direction of cycle change of rising, basically unchanged, or falling, respectively; when the previous cycle is invalid, the direction of cycle change is not reconstructed by interpolation.

[0085] Then, the processor classifies the current state into a positive slope difference state or a negative slope difference state based on whether the slope difference is greater than zero; it classifies the current state into an increasing, basically unchanged, or decreasing state based on the comparison between the current period partition temperature difference value and the previous effective period partition temperature difference value; and it classifies the current state into the corresponding range position state based on whether the current partition temperature difference value is higher than the upper limit of fluctuation, between the upper and lower limits of fluctuation, or lower than the lower limit of fluctuation. The processor combines the above three types of states into the current combined state, and uses the 95th percentile of the residual temperature rise rate in the historical effective suppression period that is the same as the current combined state as the current dynamic temperature rise rate threshold; the insufficient historical effective samples refer to the fact that the number of historical effective samples for the corresponding combined state has not reached the minimum number of samples. The minimum number of samples is the minimum number of samples when the absolute difference between two consecutive update values ​​of the 95th percentile of the combined state is not greater than the residual temperature rise rate resolution. The residual temperature rise rate resolution is determined by the ratio of the temperature measurement resolution to the measurement window duration. When there are insufficient historical valid samples for the corresponding combination state, the dynamic heating rate threshold is not updated and the previous valid value is used (for example, the dynamic heating rate threshold corresponding to the current combination state is 1.20℃ / s); when the previous valid dynamic heating rate threshold has not yet been formed, the reference sampling stage is extended, and the current cycle dynamic heating rate threshold is marked as invalid, and the continuous over-limit judgment of the current cycle is not performed. The calculation relationship of the dynamic heating rate threshold is expressed by equation (2):

[0086] (2)

[0087] In equation (2), For the first The difference between the temperature rise slope of the central stress zone and the temperature rise slope of the edge overflow zone in each pressing cycle, in °C / s; and These are the temperature differences between the current cycle and the previous effective cycle, in °C. and These are the upper and lower limits of the baseline temperature difference fluctuation, respectively, in °C. According to the pre-configured combination state division rules, the segmentation mapping rules of the 95th percentile of the residual temperature rise rate are extracted from the corresponding historical effective suppression cycle.

[0088] The segmentation mapping rule is based on Is it greater than zero? Compared to The direction of periodic changes and Compared to and The position determines the current combination state, and outputs the 95th percentile of the residual temperature rise rate of the historical effective suppression cycle corresponding to this combination state; This represents the current cycle dynamic heating rate threshold, expressed in °C / s.

[0089] Equation (2) takes the difference in temperature rise slope between the two zones, the periodic change direction of the temperature difference value between the zones, and the position of the current temperature difference value relative to the fluctuation range of the reference temperature difference as the same set of inputs, and the output results are only used for subsequent comparison of residual temperature rise rate.

[0090] Using the previous example, the temperature rise slopes of the central stress zone and the edge overflow zone are 0.36℃ / s and 0.18℃ / s, respectively, with a slope difference of 0.18℃ / s. The temperature difference between the current cycle zones is higher than that of the previous effective cycle, and the cycle change direction is upward. The current cycle zone temperature difference of 0.8℃ is higher than the upper limit of the baseline temperature difference fluctuation of 0.45℃. The above inputs correspond to the combined state of a positive slope difference, an upward cycle, and a fluctuation upper limit. This combined state corresponds to the 95th percentile of the residual temperature rise rate of the historical effective pressing cycle, which is 1.20℃ / s. Therefore, 1.20℃ / s is used as the dynamic temperature rise rate threshold for the current pressing cycle.

[0091] In this second embodiment, after the dynamic heating rate threshold is generated, the spatial gradient analysis trigger point in step S200 can be further determined through the following process:

[0092] S270. Calculate the ratio of the cumulative residual temperature rise at each temperature measurement location to the measurement window duration to obtain the residual temperature rise rate.

[0093] S280. The residual temperature rise rate is compared with the dynamic temperature rise rate threshold point by point, and the temperature measurement positions in the central stress area that continuously exceed the dynamic temperature rise rate threshold are marked as spatial gradient analysis trigger points.

[0094] S290. According to the adjacent order of the measuring points, determine the target temperature measurement position in the edge overflow area that is reachable from the spatial gradient analysis trigger point along the adjacent order of the measuring points and has a clear adjacent order, and calculate the coordinate difference, temperature rise difference and response delay between the two when the maximum temperature rise is reached.

[0095] For example, steps S270 to S290 may further include the following processes:

[0096] First, the processor calls equation (1) to divide the cumulative residual temperature rise at each temperature measurement location by the corresponding measurement window duration, obtaining the residual temperature rise rate with the same unit as the dynamic heating rate threshold. When the measurement window is not completely closed or the window duration is zero, the corresponding rate is marked as invalid.

[0097] Secondly, the processor maintains an independent consecutive limit-crossing count for each temperature measurement location in the central stress zone. The number of consecutive limit-crossings is calculated by adding one to the longest consecutive suppression period of the short-term spike during the baseline sampling phase. This is used to ensure that the continuous limit-crossing condition exceeds the observed duration of the short-term spike (for example, if the longest consecutive period of the short-term spike during the baseline sampling phase is 2 effective suppression periods, the number of consecutive limit-crossings is 3 effective suppression periods). The count is incremented when the residual temperature rise rate is higher than the current cycle's dynamic temperature rise rate threshold; the count is reset when there is no limit-crossing, the data is invalid, or the cycle number is discontinuous.

[0098] Then, after reaching the consecutive limit number, the processor determines the target temperature measurement position that is reachable from the trigger point and has a clear adjacency order along the fixed center-to-edge adjacency direction in the spatial topology, and calculates the coordinate difference between the two, the temperature rise difference in the same period, and the response delay between reaching the maximum temperature rise. If multiple identical maximum temperature rise values ​​exist within the same window, the earliest occurrence of the maximum temperature rise is taken.

[0099] Using the previous example, the residual temperature rise rates of C1 in the last three effective cycles are 1.575℃ / s, 1.925℃ / s, and 2.30℃ / s, respectively, all exceeding the dynamic temperature rise rate threshold of 1.20℃ / s. Therefore, C1 is marked as the trigger point for spatial gradient analysis. The maximum temperature rise time of C1 is 2.0s, and the maximum temperature rise time of the target edge measuring point E1 is 3.5s, with a response delay of 1.5s for both. The processor simultaneously extracts C1→M1→M2→E1 as the candidate adjacency chain for subsequent double difference calculation of adjacent measuring points.

[0100] like Figure 5 As shown, further, in step S300, a coherent heat transfer path can be obtained through the following process:

[0101] S310. Calculate the residual temperature rise difference and response delay difference between adjacent measuring points according to the adjacent measuring point order from the central stress zone to the edge overflow zone.

[0102] S320, respectively remove the two types of differences whose absolute values ​​do not exceed their respective noise tolerance limits;

[0103] S330. Determine whether the two types of differences retained are both positive along the adjacent order of the measurement points. Connect adjacent measurement points that are both positive and have continuous temperature measurement time sequence as candidate node strings. When there are multiple candidate node strings, sort them in descending order according to the number of consecutive effective connections, the minimum residual temperature rise difference within the node string, and the minimum response delay difference. Select the candidate node string with the highest sorting. If the sorting results are still the same, select the candidate node string whose starting measurement point is closest to the spatial gradient analysis trigger point to obtain the coherent heat transfer path.

[0104] For example, steps S310 to S330 may further include the following processes:

[0105] First, the path construction processor calculates the values ​​for each pair of adjacent upstream and downstream measuring points in the candidate adjacency chain, following a unified direction from the central stress area to the edge overflow area. The residual temperature rise difference is calculated by subtracting the cumulative residual temperature rise of the downstream measuring point from the cumulative residual temperature rise of the upstream measuring point, and the response delay difference is calculated by subtracting the maximum temperature rise time of the upstream measuring point from the maximum temperature rise time of the downstream measuring point.

[0106] Next, the processor compares the two types of differences with their respective noise margins. The residual temperature rise difference noise margin is determined by the temperature measurement resolution and the noise band of the residual difference between adjacent measurement points in the baseline stage (e.g., the residual temperature rise difference noise margin is 0.20℃); the response delay difference noise margin is determined by the sampling interval and the location error of the maximum temperature rise time (e.g., the response delay difference noise margin is 0.10s). If either type of difference does not exceed its noise margin, the corresponding adjacent measurement points do not form candidate connections.

[0107] Then, the processor determines whether the two types of differences have the same sign and checks whether the sampling timestamps of adjacent measurement points are continuous. The two types of differences, noise margin, and timing continuity condition are expressed by equation (3):

[0108] (3)

[0109] In equation (3), For upstream measuring points near the central stress zone, The downstream adjacent measuring point is located near the edge of the overflow area; The residual temperature rise difference is the difference between the cumulative residual temperature rise at the upstream measuring point and the cumulative residual temperature rise at the downstream measuring point, expressed in °C. The response delay difference is the difference between the maximum temperature rise time at the downstream measuring point and the maximum temperature rise time at the upstream measuring point, expressed in seconds. and These are the residual temperature rise noise margin and the response delay noise margin, respectively, with units of ℃ and s; This serves as a continuous identifier for the temperature measurement sequence. This is the identifier for the candidate directed connection.

[0110] The candidate directed connection identifier takes a valid value only when the residual temperature rise difference is greater than the residual temperature rise difference noise tolerance, the response delay difference is greater than the response delay difference noise tolerance, and the temperature measurement sequence is continuous; when any difference is negative, no candidate directed connection is formed from the central force area to the edge overflow area.

[0111] Adjacent measurement points satisfying equation (3) are connected in their original adjacency order to form candidate node strings. Candidate node strings do not cross invalid measurement points or topological breakpoints to add new connections. When only one candidate node string is formed, the candidate node string is determined as a coherent heat transfer path. When multiple candidate node strings are formed, the processor first sorts them in descending order according to the number of consecutive effective connections contained in each candidate node string, then sorts them in descending order according to the minimum residual temperature rise difference within each candidate node string, and then sorts them in descending order according to the minimum response delay difference. If the aforementioned sorting results are still the same, the candidate node string whose starting measurement point is closest to the spatial gradient analysis trigger point is selected as the coherent heat transfer path.

[0112] Continuing with the previous example, such as Figure 9 The cumulative residual temperature rise of C1, M1, M2, and E1 is 9.2℃, 8.4℃, 7.6℃, and 6.9℃, respectively. The maximum temperature rise times are 2.0s, 2.4s, 2.9s, and 3.5s, respectively. The residual temperature rise differences between adjacent values ​​are 0.8℃, 0.8℃, and 0.7℃, respectively, and the response delay differences between adjacent values ​​are 0.4s, 0.5s, and 0.6s, respectively. All of these differences exceed their respective noise margins and are positive. The temperature measurement sequence is continuous, forming... Candidate node string. This candidate node string contains three consecutive valid connections, and currently no other candidate node string with more consecutive valid connections exists, therefore it will be... It was determined to be a continuous heat transfer path.

[0113] In this second embodiment, after obtaining a coherent heat transfer path, the radial heat transfer characterization parameters in step S400 can be further formed through the following process:

[0114] S410. Calculate the ratio of the temperature change rate of the spatial gradient analysis trigger point to the temperature measurement position of the corresponding edge overflow area within the measurement window to obtain the initial rate ratio.

[0115] S420. Obtain the path direction vector of the continuous heat transfer path and the radial direction vector from the spatial gradient analysis trigger point to the temperature measurement position of the edge overflow area, and determine the directional projection coefficient based on the two direction vectors.

[0116] S430. The product of the initial rate ratio and the directional projection coefficient is used as the radial correction ratio, and the radial correction ratio is corrected according to the distance between the measuring points of the two temperature measuring positions to obtain the radial heat transfer characterization parameters.

[0117] For example, steps S410 to S430 may further include the following processes:

[0118] First, the radial analysis processor fits the temperature change rates of the spatial gradient analysis trigger point and the corresponding edge overflow area temperature measurement position within the current measurement window. The initial rate ratio is obtained by dividing the absolute value of the temperature change rate at the edge overflow area measurement position by the sum of the absolute value of the temperature change rate at the spatial gradient analysis trigger point and the rate zero division protection value. To avoid ratio distortion due to excessively small trigger point temperature change rates, the rate zero division protection value is determined by the minimum resolvable temperature change rate of the temperature measurement channel (e.g., a rate zero division protection value of 0.01℃ / s).

[0119] Secondly, the processor calculates the unit displacement vector from each adjacent upstream measuring point to the downstream measuring point according to the adjacent order of the measuring points along the continuous heat transfer path. It then calculates the arithmetic mean of each unit displacement vector and normalizes the resulting average vector to form a path direction vector. Based on spatial gradient analysis, it forms a radial direction vector from the coordinates of the trigger point to the temperature measuring position at the target edge, and normalizes this radial direction vector. The two resulting unit direction vectors are then multiplied by a dot product; if the dot product is negative, it is set to zero, yielding the non-negative direction projection coefficient.

[0120] Then, the processor uses the median of the nominal spacing of the measurement points in the same mold as the reference distance (for example, the reference distance is 10mm), and calculates the actual measurement point spacing using the coordinates of the trigger point and the temperature measurement position at the target edge. The initial rate ratio, directional projection coefficient, and distance correction relationship are expressed by equation (4):

[0121] (4)

[0122] In equation (4), and These represent the rate of temperature change within the measurement window at the temperature measurement location in the edge overflow area and the trigger point for spatial gradient analysis, respectively, in °C / s. This is the rate-to-zero protection value, in °C / s. The unit direction vector for a continuous heat transfer path; The unit radial direction vector pointing from the spatial gradient analysis trigger point to the temperature measurement position at the target edge; For reference distance, the unit is mm; The actual distance between the trigger point and the temperature measurement position at the edge of the target is expressed in mm. is a dimensionless parameter characterizing radial heat transfer. Equation (4) sequentially performs temperature change rate ratio, directional projection, and distance correction, and its output is saved in the order of path segment and window for subsequent continuous decrease judgment.

[0123] It should be noted that the path segment referred to in this embodiment refers to a complete and continuous path with fixed start and end points and a fixed order of adjacency between the same spatial gradient analysis trigger point and the corresponding temperature measurement position in the overflow area at the edge. It does not refer to a single connection between two adjacent measurement points. Between different sliding windows, only the radial heat transfer characterization parameters corresponding to the path segments with the same start and end points and the same order of adjacency are continuously compared.

[0124] Using the example above, the temperature change rates of C1 and E1 are 0.36℃ / s and 0.18℃ / s, respectively, and the initial rate ratio is about 0.486. The directional projection coefficient between the continuous path direction and the radial direction from C1 to E1 is 0.90, the reference distance is 10mm, and the actual measurement point spacing is 12mm. The radial heat transfer characterization parameter obtained from equation (4) is about 0.365.

[0125] Subsequently, as Figure 6 As shown, the suspected conduction blockage region in step S400 can be further determined through the following process:

[0126] S440. Calculate the changes in radial heat transfer characterization parameters for each path segment of the continuous heat transfer path using a sliding window.

[0127] S450. When the radial heat transfer characterization parameters of the same path segment decrease in multiple consecutive sliding windows, and the cumulative decrease exceeds the decrease threshold determined according to the baseline sampling stage, the path segment is marked as an abnormal segment.

[0128] S460. The coordinates of the abnormal section, the temperature difference between the sections, and the cumulative residual temperature rise are associated. After normalizing the three types of data, a spatial feature matrix is ​​constructed. The spatial feature matrix is ​​then processed using a mean-based clustering algorithm to obtain the spatial distribution of the suspected conduction blockage area.

[0129] For example, steps S440 to S460 may further include the following processes:

[0130] First, the continuous analysis processor saves radial heat transfer characterization parameters according to the same path segment identifier and time sequence. The path segment identifier is composed of the spatial gradient analysis trigger point identifier, the corresponding edge overflow area temperature measurement position identifier, and the fixed measurement point adjacency order, and a sliding window is formed using a fixed length and fixed step size. The sliding window length is taken as the longest consecutive window number of short-term parameter fluctuations in the same path segment during the baseline sampling phase plus one (for example, when the longest consecutive short-term parameter fluctuation exists for 3 windows, the sliding window length is 4 consecutive valid windows). To ensure that adjacent windows continuously cover the same path segment, the sliding step size is fixed at 1 window. When there are invalid parameters within a window, that window does not participate in the change calculation.

[0131] Secondly, the reduction threshold is taken as the 95th percentile of the absolute change in radial heat transfer characterization parameters between the first and last windows of the same path segment within a continuous effective window of the same length as the current sliding window during the baseline sampling phase (for example, a reduction threshold of 0.06). The same path segment is only marked as an abnormal segment if the reduction occurs window by window within a continuous sliding window and the cumulative reduction exceeds the reduction threshold; the current continuous reduction segment terminates when any window shows a rebound, becomes invalid, or is missing.

[0132] Then, the spatial analysis processor extracts the coordinates, temperature difference values, and cumulative residual temperature rise of the anomalous sections, and normalizes them according to the effective minimum and maximum values ​​of the baseline stage to form spatial feature vectors for the same temperature measurement location. When the normalization baseline range is zero, the corresponding features do not participate in clustering. The processor first divides the anomalous points into mutually independent connected candidate sets based on spatial adjacency, then calculates the arithmetic mean of the spatial feature vectors within each connected candidate set as the set center, uses Euclidean distance to calculate the feature distance between each anomalous point and the corresponding set center, and uses the 95th percentile of the feature distance within the set as the merging boundary within the set; anomalous points whose feature distance does not exceed the merging boundary and are spatially connected are merged into the same suspected conduction blockage region, while anomalous points whose feature distance exceeds the merging boundary are retained as point-level suspected locations. When there is only one anomalous point in a connected candidate set, it is directly retained as a point-level suspected location.

[0133] Following the previous example, the radial heat transfer characterization parameters for the same path segment within four consecutive effective sliding windows are 0.430, 0.410, 0.390, and 0.365, respectively. These parameters decrease in each adjacent window, with a cumulative decrease of 0.065, exceeding the decrease threshold of 0.06. Therefore, this path segment is marked as an abnormal segment. If the normalized coordinates, zone temperature differences, and cumulative residual temperature rise of M1, M2, and adjacent grid positions are close to each other and spatially connected, they are aggregated into the same suspected conduction hindrance region.

[0134] Finally, in step S500, the key monitoring locations can be further determined through the following process:

[0135] S510. Calculate the difference in residual temperature rise between the suspected conduction blockage area and the adjacent normal area within the same suppression cycle. When the difference exceeds the accumulation threshold in multiple consecutive suppression cycles, generate a continuous heat accumulation indicator.

[0136] S520. Smooth the temperature response curve of the suspected conduction blockage region and calculate the slope sequence;

[0137] S530. When the slope sequence is continuously in the preset near-zero interval for a duration exceeding the plateau duration threshold, it is determined that the temperature response curve has a plateau segment, and the position corresponding to the plateau segment is determined as the key monitoring position.

[0138] For example, steps S510 to S530 may further include the following processes:

[0139] First, the results confirm that the processor identifies a suspected conduction blockage region. Effective regions in the same radial direction or spatially adjacent regions that are not marked as abnormal are selected as adjacent normal regions. The difference in the cumulative residual temperature rise between the two regions is calculated within the same suppression cycle. The number of consecutive accumulation cycles is taken as the longest consecutive suppression cycle of short-term thermal fluctuations in the baseline sampling phase plus one (e.g., if the short-term thermal fluctuations have a maximum of 2 consecutive effective suppression cycles, the number of consecutive accumulation cycles is 3 effective suppression cycles). The accumulation threshold is determined by the high quantile of the residual difference between the suspected region and the adjacent normal region in the baseline phase (e.g., the accumulation threshold is 1.5℃). If no effective adjacent normal region exists, distant regions are not forcibly selected.

[0140] Secondly, the processor reads the temperature response curves of each effective temperature measurement location within the suspected conduction blockage area marked with continuous heat accumulation, and smooths the temperature response curves of each effective temperature measurement location. The smoothing window length is determined by the sampling interval and the platform duration threshold. An odd number of sampling points with a smoothing time span not exceeding half of the platform duration threshold is selected to avoid the smoothing process covering the entire platform interval (for example, when the sampling interval is 0.1s and the platform duration threshold is 1.2s, the smoothing window length is 5 sampling points, corresponding to 0.5s). Subsequently, the slope is fitted to the smoothed continuous local intervals corresponding to each effective temperature measurement location, forming a slope sequence corresponding to each temperature measurement location. Intervals that reach the upper limit of the temperature measurement channel, have missing samples, or maintain the same discrete value for a long time and have a saturation indicator are not included in the platform judgment.

[0141] Then, the near-zero slope boundary is determined by the temperature measurement resolution and the slope fitting error (for example, the near-zero slope boundary is 0.02℃ / s). The platform duration threshold is the sum of the 95th percentile of the continuous duration of the near-zero slope in the normal region during the benchmark sampling phase and a sampling interval, so that the duration of the platform to be confirmed exceeds the range of near-zero slope duration that has appeared in the normal region (for example, when the 95th percentile of the continuous duration of the near-zero slope in the benchmark normal region is 1.1s and the sampling interval is 0.1s, the platform duration threshold is 1.2s).

[0142] Subsequently, the processor checks whether the difference in residual temperature rise between the suspected conduction blockage area and the adjacent normal area exceeds 1.5℃ within three consecutive effective suppression cycles, and calculates the actual duration for which the absolute value of the slope of the smooth response curve at each effective temperature measurement location within the suspected conduction blockage area is continuously no greater than 0.02℃ / s. Only when the corresponding path segment has formed an abnormal segment marker, the corresponding suspected conduction blockage area has formed a continuous heat accumulation marker, and the platform duration at a certain effective temperature measurement location exceeds the platform duration threshold, are the coordinates of that effective temperature measurement location determined as a key monitoring location.

[0143] The sliding window length, decreasing threshold, number of consecutive accumulation cycles, accumulation threshold, near-zero slope boundary, and platform duration threshold used in the above continuous judgment are all given their source and example values ​​at their first occurrence. The radial heat transfer characterization parameters are calculated by equation (4) and entered into the sliding window. The results of the abnormal section are then entered into the continuous heat accumulation judgment. The continuous heat accumulation results are finally used in conjunction with the platform duration conditions, and the results of the pre- and post-processing are sequentially connected.

[0144] Continuing with the previous example, such as Figure 10 As shown, the cumulative residual temperature rise difference between the suspected conduction blockage area and the adjacent normal area over three consecutive effective suppression cycles was 1.6℃, 1.7℃, and 1.8℃, respectively, all exceeding the accumulation threshold of 1.5℃, thus indicating continuous heat accumulation. In the smooth response curve of this suspected area, a continuous interval with an absolute slope of no more than 0.02℃ / s lasted for 1.5s, exceeding the plateau duration threshold of 1.2s. Therefore, it was determined that a plateau segment existed in the temperature response curve, and the M2 neighborhood associated with the plateau segment was identified as a key monitoring location.

[0145] Once the key monitoring locations are determined, the online measurement processor outputs the online measurement results of heat conduction hindrance. The output includes the current cycle number, spatial gradient analysis trigger point, continuous heat transfer path, radial heat transfer characterization parameters, abnormal sections, suspected conduction hindrance areas, key monitoring locations, valid status, and cause of formation. Valid results are written to the historical record for sequence correlation in subsequent cycles; results awaiting verification or invalid results do not update the reference temperature difference fluctuation range, dynamic heating rate threshold, or continuity count.

[0146] During the above processing, if the measurement window is incomplete, temperature samples are missing, timestamps are discontinuous, the correspondence between the measurement point and the area fails, or the temperature value exceeds the range of the temperature measurement channel, the processor will record the corresponding measurement point-cycle as invalid, will not calculate the temperature rise, will not update the residual temperature rise accumulation, and will reset the continuous over-limit, continuous decrease, and continuous accumulation counts associated with the record.

[0147] When the reference sample is insufficient, the Kalman filter is not yet stable, the number of effective measurement points for Kriging interpolation is insufficient, the historical effective sample corresponding to the current combination state is insufficient and the previous effective dynamic heating rate threshold has not yet been formed, or a continuous candidate node string satisfying equation (3) has not been formed, the processor retains the original effective temperature record and the previous effective result, and does not generate a new key monitoring location based on the current invalid result.

[0148] When the measurement window duration is zero, the absolute value of the temperature change rate at the trigger point is not higher than the rate zero excluding the protection value, the actual measurement point distance between the trigger point and the target edge temperature measurement position is zero, the magnitude of the average vector of the path direction is zero, the position coordinates are invalid, or the direction projection coefficient is not greater than zero, the corresponding radial heat transfer characterization parameter is marked as invalid and does not enter the sliding window with a zero value.

[0149] When communication is interrupted or the result writing fails, the synchronous temperature acquisition unit or online measurement processor caches temperature records and derived states according to the cycle number; after communication is restored, the data is retransmitted according to the cycle number and integrity verification result. Data on both sides of the missing cycle is not concatenated as evidence of continuous exceeding limits, continuous decrease, or continuous accumulation.

[0150] For example, if there are adjacent measurement points in the candidate adjacent chain of the current cycle with residual temperature rise difference or response delay difference that does not exceed the corresponding noise tolerance, difference that is not positive, or temperature measurement timing interruption, the processor records the invalid path state and stops the calculation of radial heat transfer characterization parameters and key monitoring locations for the current cycle; if a continuous candidate node string that satisfies equation (3) is formed again in the next cycle, the subsequent continuous window is re-established from the effective cycle.

[0151] Accordingly, the present invention sequentially forms a zoned temperature rise residual sequence, a dynamic temperature rise rate threshold, a coherent heat transfer path, and radial heat transfer characterization parameters. It also determines key monitoring locations by combining the continuous decrease of parameters, continuous heat accumulation, and the continuous condition of a near-zero slope platform, and outputs online measurement results of heat conduction blockage.

[0152] The above description is merely a preferred embodiment of the technical solution of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for online determination of residual temperature rise in candy compression, characterized in that, include: Within the candy pressing cycle measurement window, the start and end temperatures and coordinates of the temperature measurement positions in the center stress area and the edge overflow area of ​​the mold are collected. The temperature rise at the temperature measurement position and the temperature rise difference between the two zones are calculated. The temperature rise is accumulated periodically to generate a zone temperature rise residual sequence. Based on the temperature difference fluctuation range of the two zones, the slope difference of the response curve, and the direction of periodic change during the benchmark sampling phase, a dynamic heating rate threshold is generated. The temperature measurement location of the central stress zone that continuously exceeds the ratio of the residual temperature rise to the measurement window duration is marked as the spatial gradient analysis trigger point. According to the adjacent order of the measuring points from the central stress area to the edge overflow area, calculate the residual temperature rise difference between adjacent measuring points and the response delay difference between the maximum temperature rise time. Connect the measuring points with both types of differences being positive and continuous along the adjacent order of the measuring points and with continuous temperature measurement time to obtain a coherent heat transfer path. Based on the path direction of the continuous heat transfer path, the radial direction from the spatial gradient analysis trigger point to the temperature measurement position of the edge overflow area, and the distance between the measurement points, a radial heat transfer characterization parameter is formed. The path segment where the parameter continuously decreases within multiple consecutive sliding windows and the cumulative decrease exceeds the decrease threshold is identified as a suspected conduction blockage area. When the cumulative residual temperature rise in the suspected conduction blockage area is higher than that in the adjacent normal area for multiple consecutive suppression cycles, and the slope of its response curve remains in the preset near-zero range for a duration exceeding the platform duration threshold, the corresponding location is identified as a key monitoring location, and the online measurement results of heat conduction blockage are output.

2. The method according to claim 1, characterized in that, The generated partition temperature rise residual sequence includes: The temperature rise at each temperature measurement location within the measurement window is calculated by separately calculating the difference between the end temperature and the start temperature in the central stress zone and the edge overflow zone. Calculate the temperature rise difference between the corresponding temperature measurement locations in the two zones to obtain the temperature difference value between the zones; The temperature rise at the same temperature measurement location is accumulated according to the pressing cycle to obtain the residual temperature rise accumulation. The residual temperature rise accumulation, the temperature difference value of the zone, and the coordinates are associated according to the pressing cycle to obtain the residual temperature rise sequence of the zone.

3. The method according to claim 1, characterized in that, Before generating the dynamic heating rate threshold, the method further includes correcting the residual temperature rise sequence of the partition, the correction including: Kalman filtering is used to process continuous temperature values ​​at the same temperature measurement location and synchronous temperature values ​​at adjacent measurement points to obtain smoothed temperature values ​​and smoothed synchronous values. Calculate the reference deviation between the smoothed temperature value and the smoothed synchronization value. When the reference deviation exceeds the deviation threshold, use the smoothed synchronization value to correct the smoothed temperature value at the corresponding temperature measurement position to obtain the target temperature value. The temperature rise and residual temperature rise accumulation at the corresponding temperature measurement location are recalculated based on the target temperature value. A spatial topology map is established based on the coordinates of each temperature measurement location and the adjacency relationship of the measurement points. Kriging interpolation is used to perform spatial gridding compensation on the residual temperature rise accumulation.

4. The method according to claim 1, characterized in that, The process of obtaining the temperature difference fluctuation range between the two zones includes: The temperature rise difference between the central stress zone and the edge overflow zone is calculated to form a benchmark sampling temperature difference sequence; Remove outliers from the baseline sampled temperature difference sequence and calculate the upper and lower limits of the remaining temperature difference values; The upper and lower limits are used to form a temperature rise rate threshold calculation benchmark corresponding to the benchmark sampling stage.

5. The method according to claim 1, characterized in that, The threshold for generating the dynamic heating rate includes: The temperature response curves of the central stress zone and the edge overflow zone during the current pressing cycle are respectively fitted to obtain the first temperature rise slope and the second temperature rise slope. Calculate the difference between the first temperature rise slope and the second temperature rise slope, and perform differential analysis on the temperature difference sequence of the continuous pressing cycle to obtain the direction of periodic change; The combination states are divided according to whether the slope difference is greater than zero, the direction of the period change, and the position of the current partition temperature difference value relative to the upper and lower limits of the temperature difference fluctuation range of the two partitions. The 95th percentile of the residual temperature rise rate in the historical effective suppression cycle that is the same as the current combination state is determined as the dynamic temperature rise rate threshold corresponding to the current suppression cycle. When the historical effective samples of the corresponding combination state are insufficient, the previous effective dynamic temperature rise rate threshold is used.

6. The method according to claim 1, characterized in that, The marker spatial gradient analysis trigger point includes: The residual temperature rise rate is obtained by calculating the ratio of the cumulative residual temperature rise at each temperature measurement location to the measurement window duration. The residual temperature rise rate is compared with the dynamic temperature rise rate threshold point by point, and the temperature measurement positions in the central stress area that continuously exceed the dynamic temperature rise rate threshold are marked as spatial gradient analysis trigger points. According to the adjacent order of the measuring points, the target temperature measuring position that is reachable from the spatial gradient analysis trigger point and has a clear adjacent order is determined in the edge overflow area, and the coordinate difference, temperature rise difference and response delay between the two are calculated.

7. The method according to claim 1, characterized in that, The obtained coherent heat transfer path includes: Calculate the residual temperature rise difference and response delay difference between adjacent measuring points according to the adjacent measuring point order from the central stress zone to the edge overflow zone; Remove the two types of differences whose absolute values ​​do not exceed their respective noise tolerances; Determine whether the two types of differences retained are both positive along the adjacent order of the measurement points, and connect adjacent measurement points that are both positive and have continuous temperature measurement time sequence to form a candidate node string; when there are multiple candidate node strings, sort them in descending order according to the number of consecutive effective connections, the minimum residual temperature rise difference within the node string, and the minimum response delay difference, and select the candidate node string with the highest sorting. If the sorting results are still the same, select the candidate node string that is closest to the spatial gradient analysis trigger point from the starting measurement point to obtain the coherent heat transfer path.

8. The method according to claim 1, characterized in that, The formation of radial heat transfer characterization parameters includes: Calculate the ratio of the temperature change rate of the spatial gradient analysis trigger point to the temperature measurement position of the corresponding edge overflow area within the measurement window to obtain the initial rate ratio. Obtain the path direction vector of the continuous heat transfer path and the radial direction vector from the spatial gradient analysis trigger point to the temperature measurement position of the edge overflow area, and determine the directional projection coefficient based on the two direction vectors; The product of the initial rate ratio and the directional projection coefficient is used as the radial correction ratio. The radial correction ratio is then corrected for distance based on the distance between the measuring points at the two temperature measurement locations to obtain the radial heat transfer characterization parameters.

9. The method according to claim 1, characterized in that, Identify suspected conduction block areas, including: The variation of radial heat transfer characterization parameters in each path segment of the continuous heat transfer path is calculated using a sliding window. When the radial heat transfer characterization parameters of the same path segment decrease in multiple consecutive sliding windows, and the cumulative decrease exceeds the decrease threshold determined according to the baseline sampling stage, the path segment is marked as an abnormal segment. By associating the coordinates of the abnormal section, the temperature difference between the zones, and the cumulative residual temperature rise, a spatial feature matrix is ​​constructed after normalizing the three types of data. The spatial feature matrix is ​​then processed using a mean-based clustering algorithm to obtain the spatial distribution of the suspected conduction blockage area.

10. The method according to claim 1, characterized in that, The step of identifying the corresponding location as a key monitoring location includes: Calculate the difference in residual temperature rise between the suspected conduction blockage area and the adjacent normal area within the same suppression cycle. When the difference exceeds the accumulation threshold in multiple consecutive suppression cycles, a continuous heat accumulation indicator is generated. Smooth the temperature response curve of the suspected conduction blockage region and calculate the slope sequence; When the slope sequence remains in the preset near-zero interval for a duration exceeding the plateau duration threshold, it is determined that the temperature response curve has a plateau segment, and the location corresponding to the plateau segment is identified as a key monitoring location.

Images