Ebonite rubber roller production method based on temperature and pressure intelligent regulation

Abstract

The application discloses a production method of Ebonite rubber roller based on temperature and pressure intelligent regulation and control, and relates to the technical field of rubber roller production, and comprises the following steps: constructing a temperature collection credibility function according to a local temperature abnormal distribution matrix, fitting and comparing a temperature response of each collection point with a target temperature rising curve, generating a credibility index for judging the credibility of temperature data, and correcting the judgment logic of temperature collection according to the local temperature abnormal change caused by abnormal heat conduction due to air block in the vulcanization temperature rising process. The application solves the problem of temperature control misjudgment caused by local heat conduction abnormality due to air block in the vulcanization temperature rising process of the Ebonite rubber roller, realizes dynamic identification and self-adaptive regulation and control of temperature and pressure abnormality, ensures sufficient vulcanization, uniform structure and stable performance of each area of the rubber roller, and improves product quality and service reliability.

Description

Technical Field

[0001] This invention relates to the field of rubber roller manufacturing technology, and specifically to an Ebonite rubber roller manufacturing method based on intelligent temperature and pressure control. Background Technology

[0002] Ebonite rubber roller production based on intelligent temperature and pressure control is a novel manufacturing method that integrates sensor monitoring, intelligent control algorithms, and rubber vulcanization process optimization. Its core objective is to achieve dynamic, precise, and closed-loop regulation of the two key parameters, temperature and pressure, during the Ebonite rubber roller production process, thereby improving the mechanical properties, structural density, and dimensional stability of the rubber roller. Existing Ebonite rubber roller production technologies based on intelligent temperature and pressure control primarily involve deploying temperature and pressure sensors in the vulcanization equipment to collect real-time temperature and pressure data of the rubber roller's interior and surface. This data is then transmitted to the industrial control system, where it is analyzed and judged using an embedded PID controller, fuzzy control algorithm, or a preset vulcanization model. Based on a preset process curve, the heating system (such as electric hot plates or steam heating) and pressurization system (such as hydraulic mechanisms or pneumatic devices) are dynamically adjusted to ensure uniform temperature distribution and moderate, stable pressure during vulcanization. This allows for full cross-linking of rubber molecules and effective removal of air bubbles, avoiding defects common in traditional processes such as under-vulcanization or over-vulcanization, internal voids, and uneven hardness. The entire process typically includes a preheating and pre-pressing stage, a heating and pressurizing stage, a constant temperature and pressure vulcanization stage, a cooling and depressurization stage, and a final demolding and cooling stage. Each stage is carried out under the coordinated regulation of an intelligent control system, thereby achieving high consistency, low energy consumption, and automated production of Ebonite rubber roller high-performance products.

[0003] The existing technology has the following shortcomings: In the production of Ebonite rubber rollers based on intelligent temperature and pressure control, during the vulcanization heating stage, minor deformation of the mold cavity or insufficient rubber venting may cause localized air resistance between the rubber and the mold contact surface. This leads to reduced heat conduction efficiency in the affected area, resulting in localized heating lag. Since this "thermal lag area" is not caused by insufficient heating power but by abnormal heat conduction paths, existing Ebonite rubber roller production technology based on intelligent temperature and pressure control cannot correct the abnormal temperature changes caused by air resistance during vulcanization heating to reflect these localized temperature anomalies. Consequently, it may mistakenly judge the overall temperature as adequate and prematurely enter the constant temperature control stage. This control error results in incomplete vulcanization in certain areas, ultimately leading to insufficient cross-linking of the rubber roller, creating structural weaknesses. During service, this can easily cause fatigue damage, uneven strength, or interlayer delamination, severely impacting product quality and service life.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a method for producing Ebonite rubber rollers based on intelligent temperature and pressure control, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method for producing Ebonite rubber rollers based on intelligent temperature and pressure control, specifically including the following steps: S1. Collect time-series temperature data and pressure response data at different spatial positions between the Ebonite rubber roller and the mold, and construct a thermal behavior feature field containing temperature gradient change characteristics and pressure phase characteristics to determine whether there is gas resistance causing abnormal heat conduction during the vulcanization heating process. S2. In the case of abnormal heat conduction caused by air resistance, a difference mapping model is established based on the thermal behavior characteristic field to calculate the temperature waveform delay, thermal diffusion curvature and temperature rise slope offset, and generate a local temperature anomalous distribution matrix to determine the local temperature anomalous changes in the case of abnormal heat conduction caused by air resistance during the sulfurization heating process. S3. Construct a temperature acquisition reliability function based on the local temperature anomalous distribution matrix, fit and compare the temperature response of each acquisition point with the target heating curve, generate a reliability index for judging the reliability of temperature data, and use it to correct the judgment logic of temperature acquisition based on the local temperature anomalous changes caused by gas resistance during the sulfurization heating process. S4. Reconstruct the judgment logic of temperature acquisition based on the reliability index of temperature acquisition, adopt a multi-parameter joint judgment structure to classify and identify the heating state, and adjust the temperature judgment node, heating duration and heat input distribution scheme. S5. Calculate the temperature and pressure balance coefficient based on the judgment logic correction result, and compare the temperature output and pressure response synchronously during the heating stage to execute a real-time temperature and pressure adjustment strategy to achieve dynamic control of the heating process.

[0007] Preferably, S1 is as follows: Temperature and pressure acquisition units are arranged radially and axially in the contact area between the Ebonite rubber roller and the mold. Temperature and pressure response data at each spatial location are acquired synchronously by setting a time series sampling period. The correspondence between sampling points and spatial coordinates is established in time order to form a time series temperature and pressure response dataset for different spatial locations. The time-series temperature data and pressure response data are matched and calculated according to spatial coordinates. By calculating the rate of change of temperature gradient and pressure phase difference between adjacent acquisition units, a multi-dimensional parameter matrix containing temperature gradient change characteristics and pressure phase characteristics is constructed. This multi-dimensional parameter matrix is ​​then spatially mapped to form a thermal behavior feature field, which is used to reflect the thermal and pressure coupling relationship between the Ebonite rubber roller and the mold. By jointly comparing and analyzing the temperature gradient change characteristic parameters and pressure phase characteristic parameters in the thermal behavior characteristic field, when a discontinuous abrupt change in the temperature gradient change characteristic and a synchronous shift in the pressure phase characteristic are detected, it is determined that there is an abnormal situation of heat conduction caused by gas resistance during the sulfurization heating process.

[0008] Preferably, S2 specifically includes the following steps: S201. In the case of abnormal heat conduction caused by air resistance, a difference mapping model is constructed based on the thermal behavior feature field. By comparing the temperature gradient change characteristics and pressure phase characteristics of the normal and abnormal regions in the thermal behavior feature field under the same spatial path, the spatial response difference parameter is extracted and mapped to a unified thermal behavior reference grid to reflect the heat conduction path change characteristics under air resistance interference. S202. Based on the difference mapping model, the time series offset analysis method is used to analyze the temperature change curve of each spatial node in the thermal behavior feature field, calculate the temperature waveform delay to identify the heat conduction time lag, calculate the heat diffusion curvature based on the three-point method to identify the heat flow return or focusing area, and obtain the temperature rise slope offset through differential calculation to quantify the local heating abnormal trend. S203. The temperature waveform delay, thermal diffusion curvature and temperature rise slope offset are constructed into a local temperature anomalous distribution matrix according to the spatial node order. By matching the results of multi-index coupling in the matrix with the threshold, the local temperature anomalous change region under the condition of abnormal heat conduction caused by gas resistance during the sulfurization heating process is identified, and the spatial coordinates and thermal feature labels of the corresponding region are output.

[0009] Preferably, S202 specifically refers to: Based on the difference mapping model, the temperature change curve of each spatial node in the thermal behavior feature field is time normalized. By setting a uniform sampling time interval, the temperature change sequences of different nodes are aligned with time index to form a temperature sequence set with a synchronous time reference for subsequent time series offset analysis. In the temperature sequence set, temperature change curve pairs of adjacent spatial nodes are selected. The time offset between each curve is calculated by sliding window to obtain the temperature waveform delay parameter. The numerical change rate of adjacent temperature points is fitted by the three-point method to calculate the thermal diffusion curvature, thereby obtaining the local curvature distribution of the heat flow direction. Differential operation is performed on the time-continuous sampled values ​​of each node in the temperature sequence set to obtain the change in temperature rise rate and couple it with the thermal diffusion curvature data to extract the temperature rise slope offset to characterize the local heating trend change features, and output it in the form of a parameter group for subsequent matrix generation process.

[0010] Preferably, S3 specifically includes the following steps: S301. Based on the temperature waveform delay, thermal diffusion curvature and temperature rise slope offset of each spatial node in the local temperature anomaly distribution matrix, construct a three-dimensional thermal anomaly feature vector set, set the confidence weights corresponding to each feature dimension, and generate node-level thermal behavior perturbation values ​​through vector weighting calculation, which are used to construct the temperature acquisition confidence function. S302. Compare the thermal behavior perturbation value output by the temperature acquisition confidence function with the temperature response data of each acquisition point, and perform curve fitting between the temperature response data and the target heating curve based on the sliding window method. Calculate the fitting residual and response delay difference index to quantify the degree of thermal response offset of the acquisition point. S303. The thermal response offset degree and the temperature acquisition credibility function result are weighted and fused to generate a credibility index for judging the credibility of temperature data. The judgment logic of temperature acquisition is corrected according to the credibility index, and the acceptance level of acquisition points, judgment trigger conditions and time fault tolerance range are adjusted to realize the responsive identification of local abnormal temperature changes and the optimization of acquisition strategy.

[0011] Preferably, S302 is as follows: The thermal behavior perturbation values ​​output by the temperature acquisition confidence function are paired and compared with the temperature response data corresponding to each acquisition point in spatial coordinate order. By establishing a time synchronization index table, a one-to-one correspondence between the thermal behavior perturbation values ​​and the temperature response data of each acquisition point is realized, forming a matching dataset, which provides synchronous data input for curve fitting analysis. In the matching dataset, select temperature response data segments within a continuous time window, apply the sliding window method to slide sequentially with a set window width and step size, and use the least squares fitting algorithm to fit the temperature response data with the target heating curve. Record the fitting residual value and response delay difference value at each window position to capture the local heating response shift in time. The fitting residuals and response delay differences of all sliding windows are normalized and calculated to generate thermal response offset curves that reflect the entire heating process of each sampling point. By analyzing the fluctuation frequency and amplitude of the offset curves, stable deviation intervals are extracted and a set of thermal response offset parameters is formed, providing input data basis for subsequent confidence index calculation.

[0012] Preferably, S4 is as follows: Based on the reliability index of temperature acquisition, all temperature acquisition points are classified into reliability levels, low reliability points are eliminated and a reliability index table is generated. A one-to-one mapping relationship between the index table and the distribution of thermal behavior disturbance is established, the judgment logic of temperature acquisition is reconstructed, and the reconstructed logic is used as the input premise structure of the judgment system. Based on the reconstructed judgment logic, the reliability index of temperature acquisition, the degree of thermal response offset and spatial node features are integrated to construct a joint judgment structure with multiple parameter dimensions. Density clustering and support vector classification algorithms are used to classify the heating state and output classification labels for steady state, thermal hysteresis state and nonlinear fluctuation state. The temperature acquisition judgment strategy parameters are updated based on the temperature rise status classification label. By dynamically adjusting the selection strategy and distribution density of temperature judgment nodes, the target range of the temperature rise duration is reset, and the power ratio of each heat input path is adjusted and optimized. This achieves the linkage reconstruction of the heat input distribution scheme in the temporal and spatial dimensions to adapt to the changing characteristics of the actual temperature rise status during the vulcanization process.

[0013] Preferably, S5 is as follows: Based on the temperature acquisition reliability index and thermal response offset of each spatial node in the judgment logic correction result, the temperature change gradient and pressure response gradient in the time series are jointly extracted to construct a set of temperature and pressure characteristic parameters. The temperature and pressure balance coefficient is calculated by weighted fusion method to quantify the degree of change in the coupling relationship between temperature output and pressure response during the heating stage. During the heating phase, the temperature output and pressure response are synchronously compared based on the temperature-pressure balance coefficient. By establishing a synchronization time index table, the dynamic change trajectories of temperature and pressure at each moment are compared. The sliding window algorithm is used to identify the delay alignment point and synchronization offset trend of the temperature-pressure response curve, and generate a temperature-pressure alignment offset curve for judging the thermal-pressure coordination during the heating process. Based on the synchronous offset values ​​of each time period in the temperature and pressure alignment offset curve, an adaptive adjustment algorithm is used to construct a real-time temperature and pressure adjustment strategy. The heat input power distribution and pressure loading rate during the heating stage are dynamically adjusted according to the sampling period. When the temperature and pressure coupling offset exceeds the control tolerance range, the control parameters are corrected in time to achieve synchronous temperature and pressure control and dynamic stable matching during the heating process.

[0014] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs a thermal behavior characteristic field and identifies coupling anomalies between temperature gradient change characteristics and pressure phase characteristics, enabling precise judgment of abnormal heat conduction paths caused by gas resistance during the vulcanization heating process. Based on this, the system employs difference mapping modeling and multi-parameter thermal response characteristic analysis methods to effectively identify local heating lag regions. Furthermore, it quantifies and evaluates temperature acquisition data using a confidence function, achieving the identification and differentiation of thermal anomaly disturbance regions. This corrects the problem in traditional temperature control systems where the heating process is prematurely terminated due to misjudgment of the target temperature. This response logic, built upon the laws of physical field changes, improves the decision-making reliability and control accuracy of temperature acquisition data, effectively avoiding structural defects caused by insufficient local vulcanization.

[0015] 2. This invention, by introducing intelligent judgment mechanisms such as a reliability index, a multi-parameter joint judgment structure, and a temperature-pressure balance coefficient, realizes a closed-loop optimized control system from data acquisition and judgment to state identification and control strategy execution. During the heating process, a sliding window curve fitting and thermal response offset identification method are used, which not only improves the accuracy of heating state classification but also provides a basis for the dynamic matching of subsequent heat input power and pressure loading rate. Finally, an adaptive adjustment algorithm achieves a closed-loop response for thermo-pressure coupling control, making the temperature and pressure regulation process more precise and real-time, significantly improving the stability, adaptability, and product quality consistency of the vulcanization process under complex working conditions. The overall technical solution enhances the ability to identify and control non-ideal heat transfer states, is applicable to the molding process of rubber products with various complex mold cavity structures, and has significant engineering application value. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0017] Figure 1 This is a schematic diagram of the process for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to the present invention. Detailed Implementation

[0018] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0019] This invention provides, for example Figure 1The Ebonite rubber roller production method based on intelligent temperature and pressure control shown includes the following steps: S1. Collect time-series temperature data and pressure response data at different spatial positions between the Ebonite rubber roller and the mold, and construct a thermal behavior feature field containing temperature gradient change characteristics and pressure phase characteristics to determine whether there is gas resistance causing abnormal heat conduction during the vulcanization heating process. In this embodiment, S1 specifically refers to: Temperature and pressure acquisition units are arranged radially and axially in the contact area between the Ebonite rubber roller and the mold. Temperature and pressure response data at each spatial location are acquired synchronously by setting a time series sampling period. The correspondence between sampling points and spatial coordinates is established in time order to form a time series temperature and pressure response dataset for different spatial locations. The process of arranging temperature and pressure acquisition units radially and axially in the contact area between the Ebonite rubber roller and the mold can be achieved by embedding miniature thermocouples and piezoelectric sensors at multiple specific locations on the inner wall of the mold and the outer periphery of the rubber roller. Each thermocouple continuously acquires the real-time temperature change at its location, while the piezoelectric sensor records the pressure response at that point. The sampling system simultaneously triggers all sensors at a set time series period and stores the sampled data in order of timestamp. Spatial positioning is achieved by marking the absolute position coordinates of each acquisition point in the rubber roller coordinate system during installation and embedding location information tags in the data stream. This creates a temperature and pressure data stream covering the entire contact area and different spatial locations, achieving synchronous acquisition in both time and spatial dimensions. This method allows for the acquisition of the true distribution of heat and pressure transfer between the rubber and the mold, facilitating the subsequent construction of a feature field with analytical capabilities, thus providing a basis for accurately identifying heat conduction anomalies.

[0020] The temperature acquisition unit uses miniature thermocouples or fiber optic temperature sensors as its core components, possessing high temporal resolution and high-temperature response sensitivity. It can operate stably for extended periods in the high-temperature environment of rubber vulcanization and accurately reflect minute temperature rise changes in localized areas. The pressure acquisition unit typically employs thin-film piezoresistive sensors or piezoelectric ceramic sensors, characterized by high repeatability and low nonlinearity error, enabling precise sensing of instantaneous pressure fluctuations during mold closure. These sensors are embedded between the rubber contact surface and the mold edge using customized encapsulation materials and bonding processes, ensuring that they do not interfere with the normal molding process while preventing signal distortion due to high temperatures or elastic interference. Each acquisition unit is physically address-mapped through encoding. At the signal processing end, a synchronous acquisition controller performs time alignment and spatial positioning of data from different units, thereby achieving a one-to-one correspondence between temperature data and pressure response data at the same spatial location and time point, forming a high-resolution thermo-pressure state distribution map. This acquisition configuration allows for the creation of a refined raw dataset, providing precise evidence for subsequent analysis of abnormal thermal behavior.

[0021] The time-series temperature data and pressure response data are matched and calculated according to spatial coordinates. By calculating the rate of change of temperature gradient and pressure phase difference between adjacent acquisition units, a multi-dimensional parameter matrix containing temperature gradient change characteristics and pressure phase characteristics is constructed. This multi-dimensional parameter matrix is ​​then spatially mapped to form a thermal behavior feature field, which is used to reflect the thermal and pressure coupling relationship between the Ebonite rubber roller and the mold. Matching and calculating time-series temperature and pressure response data according to spatial coordinates can be achieved through synchronous matching using spatial point coordinate indexes established before data acquisition. Each data acquisition unit is assigned a unique three-dimensional coordinate identifier during acquisition, and the system merges the time series of temperature and pressure data based on the same coordinate points during data processing. During the calculation process, paired data paths are first established between adjacent spatial acquisition units. The temperature values ​​at the same moment for each pair of acquisition points are differentiated and divided by the actual physical distance between the two points to obtain the rate of change of the temperature gradient. Similarly, the corresponding pressure response curves are compared synchronously in phase, and their phase offset is calculated and mapped to a standardized index. Taking two axial points on a rubber roller as an example, if the temperature changes rapidly at the former position and lags behind at the latter, and the corresponding pressure response is significantly earlier or later, a bidirectional feature point with a high rate of change of temperature gradient and a large pressure phase difference can be constructed. The rate of change of temperature gradient and the pressure phase difference of all adjacent sampling pairs are filled into a three-dimensional parameter matrix according to spatial topology, and spatial reconstruction is performed based on the point coordinates, ultimately forming a thermal behavior feature field to demonstrate the linkage between heat and pressure changes. This treatment method can reveal the microscopic dynamic behavior caused by uneven hot pressing or air resistance during the vulcanization process of rubber rollers.

[0022] Time-series temperature data refers to a continuously recorded numerical sequence of temperature changes, revealing the thermal dynamics trend of a sampling point during the heating process. Pressure response data refers to the pressure change trajectory at the same location, reflecting the real-time evolution of mold closure and rubber compression. Spatial coordinate matching means that temperature and pressure data have a unified reference point in a three-dimensional coordinate system, enabling precise pairing analysis. The rate of change of temperature gradient is an indicator of the strength of heat propagation in space; an abnormal increase usually indicates impeded heat diffusion. Pressure phase difference is used to characterize the temporal differences between pressure signals in adjacent areas and is an important parameter for identifying air resistance interference or local delayed responses. The multidimensional parameter matrix is ​​a set formed by structurally arranging the rate of change of temperature gradient and pressure phase difference between points, possessing directionality and spatial continuity. The thermal behavior feature field is the result of mapping the above parameter matrix to the actual spatial topology, which can be regarded as a holographic representation of the thermal and pressure coupling state between the Ebonite rubber roller and the mold, providing a quantifiable data basis for subsequent identification of abnormal areas and judgment of air resistance effects.

[0023] By jointly comparing and analyzing the temperature gradient change characteristic parameters and pressure phase characteristic parameters in the thermal behavior characteristic field, when a discontinuous abrupt change in the temperature gradient change characteristic and a synchronous shift in the pressure phase characteristic are detected, it is determined that there is an abnormal situation of heat conduction caused by gas resistance during the sulfurization heating process.

[0024] The process of jointly comparing and analyzing the temperature gradient change characteristic parameters and pressure phase characteristic parameters in the thermal behavior characteristic field can be achieved through the parameter mapping relationship of each spatial node in the thermal behavior characteristic field. First, the temperature gradient change characteristic parameters are arranged in a spatial sequence, the gradient change trend between consecutive nodes is calculated, and discontinuities such as sudden jumps or reversals in the rate of change are detected. Then, the time series curve of the pressure phase characteristic parameters is synchronously compared with the temperature gradient change trend to identify the relative offset of the two within the same spatial interval. When the temperature gradient change trend suddenly stops in a certain region, while the pressure phase curve is delayed or shifted earlier relative to the adjacent region, it indicates that the heat conduction process has been locally blocked, that is, the heat transfer path is disturbed, and it can be determined that there is a gas resistance phenomenon in this region. Taking a real production scenario as an example, during the vulcanization process of rubber rollers, when gas is trapped on the local contact surface of the mold, the temperature gradient curve in this region shows a nonlinear sudden drop, while the pressure phase curve shows an abnormal delay. The two characteristics coincide in spatial position, which indicates the formation of abnormal heat conduction. This joint comparison analysis logic can accurately identify heat conduction obstacles caused by air resistance without relying on human judgment, providing a reliable triggering basis for subsequent logic correction.

[0025] Temperature gradient variation characteristic parameters describe the changes in the diffusion rate of heat between different spatial nodes, and their continuity reflects the unobstructed nature of heat conduction paths. A sudden discontinuity in this parameter indicates that heat conduction in a certain region is blocked or delayed. Pressure phase characteristic parameters characterize the synchronicity of the pressure signal response in the time dimension, and their changes are directly related to the dynamic behavior of rubber's expansion and compression after heating. When the pressure phase characteristic shifts synchronously, it indicates that the mechanical response and thermal behavior in a local area are not synchronized, usually caused by gas barriers or conduction hysteresis. Joint comparative analysis is a computational method that integrates temperature gradient characteristics and pressure phase characteristics in both spatial and temporal dimensions. By cross-checking the coupling trends of the two sets of parameters, abnormal synergistic phenomena of heat conduction and pressure transfer can be revealed. The abnormal heat conduction caused by gas resistance is the region in this comparison where abrupt changes in the thermal field and pressure shifts coexist synchronously. The identification result of this region is directly fed back to the judgment logic correction module for dynamic adjustment of subsequent heating control strategies.

[0026] S2. In the case of abnormal heat conduction caused by air resistance, a difference mapping model is established based on the thermal behavior characteristic field to calculate the temperature waveform delay, thermal diffusion curvature and temperature rise slope offset, and generate a local temperature anomalous distribution matrix to determine the local temperature anomalous changes in the case of abnormal heat conduction caused by air resistance during the sulfurization heating process. In this embodiment, S2 specifically includes the following steps: S201. In the case of abnormal heat conduction caused by air resistance, a difference mapping model is constructed based on the thermal behavior feature field. By comparing the temperature gradient change characteristics and pressure phase characteristics of the normal and abnormal regions in the thermal behavior feature field under the same spatial path, the spatial response difference parameter is extracted and mapped to a unified thermal behavior reference grid to reflect the heat conduction path change characteristics under air resistance interference. Under the premise of air resistance causing abnormal heat conduction, the temperature and pressure responses in some regions of the thermal behavior feature field will deviate from normal heating behavior. The construction of a difference mapping model can be achieved by selecting representative normal and suspected abnormal regions in the thermal behavior feature field and comparing their temperature gradient change characteristics and pressure phase characteristics along the same spatial path point by point. Specifically, a constant-step sampling method can be used to extract the temperature change rate and pressure response phase difference at each node in the spatial path, forming a set of two-dimensional comparison sequences. By calculating the difference in characteristic parameters between the two regions along the same spatial path, such as temperature gradient abrupt change points and pressure fluctuation lag points, a set of spatial response difference parameters is obtained. Subsequently, these parameters are mapped onto a uniformly constructed thermal behavior reference grid, and a difference mapping map is formed through two-dimensional interpolation or three-dimensional thermal field reconstruction, thereby revealing the abnormal heat conduction path caused by air resistance interference, such as local countercurrent regions or heat accumulation regions deviating from the main heat flow direction. This mapping structure can intuitively reveal spatial regions of thermal coupling imbalance, providing data support for subsequent identification of local anomalies.

[0027] The thermal behavior feature field is a high-dimensional spatiotemporal field modeled by integrating the temperature gradient changes and pressure phase responses at various spatial points between the Ebonite rubber roller and the mold, reflecting the dynamic coupling behavior of heat and pressure. The temperature gradient change feature refers to the rate of change of temperature with distance along a spatial path, used to capture the spatial distribution of heat conduction efficiency. The pressure phase feature refers to the relative time shift of the pressure response at different points at the moment of heat input, indirectly reflecting the delay or lag of the rubber's internal response to changes in external pressure. The spatial response difference parameter is a difference index constructed based on the above two features, used to quantify the degree of deviation in heat conduction behavior between normal and suspected abnormal regions. The unified thermal behavior reference grid is a spatial reconstruction structure that provides structural uniformity for comparing thermal behavior in multiple regions by normalizing coordinates and parameter distribution, ensuring the comparability of difference values ​​within the same reference system. The difference mapping model not only constructs a spatial "deviation map" but also provides an analytical means to structure and visualize complex thermal anomalies, serving as a key foundation for identifying the influence of non-uniform thermal barriers.

[0028] S202. Based on the difference mapping model, the time series offset analysis method is used to analyze the temperature change curve of each spatial node in the thermal behavior feature field, calculate the temperature waveform delay to identify the heat conduction time lag, calculate the heat diffusion curvature based on the three-point method to identify the heat flow return or focusing area, and obtain the temperature rise slope offset through differential calculation to quantify the local heating abnormal trend. S203. The temperature waveform delay, thermal diffusion curvature and temperature rise slope offset are constructed into a local temperature anomalous distribution matrix according to the spatial node order. By matching the results of multi-index coupling in the matrix with the threshold, the local temperature anomalous change region under the condition of abnormal heat conduction caused by gas resistance during the sulfurization heating process is identified, and the spatial coordinates and thermal feature labels of the corresponding region are output.

[0029] The process of constructing a local temperature anomaly distribution matrix involves filling a three-dimensional data matrix point by point with three key thermal behavior parameters: temperature waveform delay, thermal diffusion curvature, and temperature rise slope offset, according to the order of their spatial nodes. Each spatial node occupies one row in the matrix, and its corresponding three columns record the waveform delay, diffusion curvature, and temperature rise slope offset during the heat conduction process, thus forming a dataset with a unified structure. To identify local temperature anomaly change regions, a set of threshold intervals based on historical production data statistics can be designed to filter the feature values ​​in each column. For example, upper limits for waveform delay, negative diffusion curvature ranges, and thresholds for sharp increases in slope offset can be set. By judging whether the three parameters of a node simultaneously meet the abnormal conditions, multi-indicator coupled anomaly identification of that node can be achieved. If a node is marked as abnormal, its spatial coordinates and the three index values ​​will be extracted and output as a set of thermal feature labels for subsequent control strategy adjustments and temperature and pressure compensation model calls. By integrating multi-dimensional thermal behavior indicators into a single matrix and performing coupled judgment, the discrimination accuracy of complex thermal barrier phenomena can be significantly improved.

[0030] The temperature waveform delay in this process is a parameter for evaluating the time lag of heat transfer, typically obtained by identifying the offset of the thermal response starting point through a sliding window. Thermal diffusion curvature describes the back-and-forth or aggregation characteristics of heat flow in spatial distribution, reflecting the perturbation effect of the structure on the heat flow. The temperature rise slope offset quantifies the anomalous trend of the temperature change rate, revealing whether the heating intensity exhibits nonlinear growth in local areas. These three features collectively characterize the anomaly of the heat conduction path from three dimensions: time, space, and dynamic change. The constructed local temperature anomaly distribution matrix not only preserves the spatial topological relationships of each node but also integrates its thermal behavior state, ensuring the consistency and traceability of the identification results in physical space. The spatial coordinate output facilitates the subsequent location of key intervention areas in the heating strategy, while the generation of thermal feature labels provides a quantitative basis for intelligent decision-making. Through this method, the thermal hysteresis distribution caused by air resistance can be systematically captured, thereby achieving early warning and intelligent intervention for local anomalies in the vulcanization process.

[0031] In this embodiment, S202 specifically refers to: Based on the difference mapping model, the temperature change curve of each spatial node in the thermal behavior feature field is time normalized. By setting a uniform sampling time interval, the temperature change sequences of different nodes are aligned with time index to form a temperature sequence set with a synchronous time reference for subsequent time series offset analysis. After the difference mapping model is constructed, the temperature change curves of each spatial node in the thermal behavior feature field need to be time-normalized to ensure that the data at different locations have a unified time reference basis. Specifically, a fixed sampling time interval can be set first, and the temperature data of all spatial nodes can be resampled to this interval. Data alignment is then performed using linear interpolation, spline interpolation, or Fourier interpolation to construct a time-consistent temperature change sequence. The time index in each sequence is strictly unified, ensuring that the temperature data of each node can be directly compared laterally at any time point. This process forms a temperature sequence set with a synchronized time reference, enabling the clear identification of the response order between nodes during subsequent time series offset analysis. For example, if the temperature rise of a node lags behind other nodes by a complete sampling period, it can be considered an initial sign of heat conduction obstruction at that location. In this process, "time normalization" is used to eliminate unstructured timing errors caused by acquisition delay, communication interference, or local thermal capacity differences; "uniform sampling time interval" ensures that all temperature sequences are comparable; and "alignment by time index" provides a timing basis for subsequent calculations of temperature waveform delay, thermal diffusion curvature, and temperature rise slope, making the entire thermal behavior analysis have strict time consistency and computability.

[0032] In the temperature sequence set, temperature change curve pairs of adjacent spatial nodes are selected. The time offset between each curve is calculated by sliding window to obtain the temperature waveform delay parameter. The numerical change rate of adjacent temperature points is fitted by the three-point method to calculate the thermal diffusion curvature, thereby obtaining the local curvature distribution of the heat flow direction. In a temperature sequence set, pairs of temperature change curves from adjacent spatial nodes are selected. These pairs are matched based on the spatial coordinate relationship of the nodes in the thermal behavior characteristic field. Then, a sliding window analysis method is applied to compare the time response differences between each pair of curves. Specifically, one reference curve is kept stationary while the other target curve is shifted point-by-point along the time dimension. By calculating the mean square error of the two curves at each position within the window, the position with the smallest error is found as the time offset point, thus obtaining the temperature waveform delay parameter for that curve pair. This parameter reflects the actual time difference of heat conduction between two adjacent nodes. Subsequently, the adjacent temperature values ​​of each node are fitted using a three-point method to calculate the temperature change rate at that node. The second derivative of the fitting results for the three adjacent nodes is then used to estimate the thermal diffusion curvature. This parameter reveals the curvature change of heat flow during spatial propagation, thereby determining whether heat is focusing, reflected, or deviating. For example, when the thermal diffusion curvature is negative and shows a continuous decreasing trend, it can be determined that heat flow is either reversing or obstructed. In this way, the sliding window provides a means of analyzing response differences at the time level, while the three-point rule captures changes in heat flow patterns at the spatial level. The combination of the two can systematically identify the dynamic behavior of heat propagation in complex structures, which helps to reveal local thermal hysteresis or energy accumulation areas caused by air resistance and has significant implications for the quantitative identification of heat conduction anomalies.

[0033] Differential operation is performed on the time-continuous sampled values ​​of each node in the temperature sequence set to obtain the change in temperature rise rate and couple it with the thermal diffusion curvature data to extract the temperature rise slope offset to characterize the local heating trend change features, and output it in the form of a parameter group for subsequent matrix generation process.

[0034] The purpose of differential calculation on the time-series continuous sampled values ​​of each node in the temperature sequence set is to quantify the rate of temperature change per unit time, i.e., to obtain the change in the rate of temperature rise. In practice, this can be achieved by subtracting the temperature value at each moment from the temperature value at the previous moment and dividing by the sampling time interval, forming a rate sequence of the same length as the original temperature sequence. This rate sequence can reveal the strength and rhythm of the response at different nodes during the heating process. Subsequently, this rate of temperature rise change is coupled point-by-point with the thermal diffusion curvature obtained in the previous calculation step. The coupling process can employ normalization fusion or weighted superposition methods, using the rate of temperature change and thermal flux curvature as two-factor indicators to construct a temperature rise slope offset. This offset reflects whether a node experiences a sudden change in the rate of temperature rise along with an abnormal thermal diffusion curvature. For example, when the thermal diffusion curvature of a node is extremely high and the rate of temperature rise increases sharply, it usually indicates an abnormal heating trend caused by localized heat energy focusing, which easily leads to the formation of localized overheating zones. By organizing the temperature rise slope offset of each node into a vector and naming it a parameter group, it is easy to directly reference it when constructing the local temperature anomaly distribution matrix, and it serves as an important basis for identifying thermal hysteresis regions. This method comprehensively considers the dynamic changes over time and the curvature of the spatial thermal field, resulting in higher accuracy in locating thermal anomalies.

[0035] S3. Construct a temperature acquisition reliability function based on the local temperature anomalous distribution matrix, fit and compare the temperature response of each acquisition point with the target heating curve, generate a reliability index for judging the reliability of temperature data, and use it to correct the judgment logic of temperature acquisition based on the local temperature anomalous changes caused by gas resistance during the sulfurization heating process. In this embodiment, S3 specifically includes the following steps: S301. Based on the temperature waveform delay, thermal diffusion curvature and temperature rise slope offset of each spatial node in the local temperature anomaly distribution matrix, construct a three-dimensional thermal anomaly feature vector set, set the confidence weights corresponding to each feature dimension, and generate node-level thermal behavior perturbation values ​​through vector weighting calculation, which are used to construct the temperature acquisition confidence function. To extract core data helpful for judging the credibility of thermal anomalies from the local temperature anomaly distribution matrix, it is first necessary to standardize the temperature waveform delay, thermal diffusion curvature, and temperature rise slope offset of each spatial node in the matrix, unifying the numerical dimensions of different physical quantities. Next, a three-dimensional thermal anomaly feature vector is constructed for each node, corresponding to the behavioral characteristics of the temperature response in three dimensions: time, spatial diffusion, and heating rate. To highlight the credibility role of different features in the judgment, credibility weights should be assigned to each dimension based on empirical data or machine learning model results; for example, the weight can be increased in cases where thermal diffusion anomalies are more sensitive. Next, a weighted vector calculation method is used to sum the inner product of the three-dimensional features and their corresponding weights to obtain the thermal behavior perturbation value for each node. This perturbation value quantifies the degree to which the node's thermal behavior may mislead the determination of the sulfidation state and can serve as the basic input for constructing the temperature acquisition credibility function. This structured processing not only preserves the three types of key anomaly information but also improves the sensitivity and stability of subsequent fitting judgments.

[0036] Temperature waveform delay reflects the time lag of the temperature response at a certain acquisition point relative to the ideal curve during the heating process, revealing an obstruction in the heat conduction path at that point. Thermal diffusion curvature measures the degree of deflection of the temperature field during spatial propagation; an abnormal curvature at a point may indicate heat convergence or dispersion in that region. Temperature rise slope offset characterizes the deviation between the heating rate at a certain point and the expected rate during heating, and is a key parameter revealing local heating efficiency imbalances. The three-dimensional thermal anomaly feature vector is a structural unit constructed using these three types of physical parameters as dimensions, used to identify the degree of thermal anomaly at each node in the spatial dimension. Feature confidence weights are a set of preset weighting coefficients, set according to the decision-making influence of each physical dimension in actual control, used to emphasize the contribution of different features to the overall confidence judgment. Node-level thermal behavior perturbation values ​​are the weighted calculation result based on the feature vector and weight vector, a quantitative evaluation of the acquisition point's potential to mislead temperature judgments, used for weight adjustment or confidence weighting when subsequently constructing the temperature acquisition confidence function.

[0037] S302. Compare the thermal behavior perturbation value output by the temperature acquisition confidence function with the temperature response data of each acquisition point, and perform curve fitting between the temperature response data and the target heating curve based on the sliding window method. Calculate the fitting residual and response delay difference index to quantify the degree of thermal response offset of the acquisition point. S303. The thermal response offset degree and the temperature acquisition credibility function result are weighted and fused to generate a credibility index for judging the credibility of temperature data. The judgment logic of temperature acquisition is corrected according to the credibility index, and the acceptance level of acquisition points, judgment trigger conditions and time fault tolerance range are adjusted to realize the responsive identification of local abnormal temperature changes and the optimization of acquisition strategy.

[0038] To achieve responsive identification and optimization of acquisition strategies for anomalous local temperature changes, it is first necessary to weight and fuse the thermal response offset degree with the temperature acquisition confidence function result. Specifically, this involves constructing a weighted fusion model, setting the weight ratios of the thermal response offset degree and thermal behavior perturbation value on different feature dimensions, and using linear weighting or Bayesian probabilistic weighting methods to merge the two to generate a confidence index. The confidence index serves as a comprehensive evaluation of the reliability of temperature data at the acquisition point; a higher value indicates lower confidence in the temperature data at that point. In practical implementation, multiple confidence threshold levels can be set, and the temperature acquisition judgment logic can be dynamically adjusted based on the confidence index. For example, when the confidence index of a certain acquisition point exceeds a set threshold, the confidence level of that point in subsequent temperature control processes is lowered, its weight in temperature control judgment is reduced, and the severity of the triggering conditions is increased, extending the time tolerance range of that point to avoid the risk of misjudgment. If multiple adjacent acquisition points simultaneously exhibit high confidence indices, a local abnormal temperature rise warning can be issued, and a local heating compensation model can be invoked to execute response actions. This process effectively identifies areas of abnormal heat conduction and corrects the acquisition strategy, enabling dynamic adaptive adjustment of the temperature data judgment logic during the heating process.

[0039] The thermal response offset reflects the temporal stability and deviation of the temperature response, and is a stability index derived from the sliding window residual curve. The temperature acquisition reliability function is a multi-dimensional weighted function constructed from spatial thermal behavior anomaly characteristics, outputting a numerical result reflecting the intensity of thermal behavior disturbances. The weighted fusion of these two functions can take into account both temporal continuity and spatial anomaly characteristics, improving the accuracy of temperature data reliability judgment. The reliability index is a single discriminant index generated after fusion, facilitating direct invocation of control logic for setting judgment thresholds and behavioral responses. The reliability level adjustment is a dynamic control mechanism for the weight of single-point data; the triggering conditions and the setting of the time tolerance range jointly determine whether the data point participates in the current stage of control decision-making. This structure possesses significant adaptability and error tolerance, representing an optimization and upgrade of traditional static temperature acquisition strategies.

[0040] In this embodiment, S302 specifically refers to: The thermal behavior perturbation values ​​output by the temperature acquisition confidence function are paired and compared with the temperature response data corresponding to each acquisition point in spatial coordinate order. By establishing a time synchronization index table, a one-to-one correspondence between the thermal behavior perturbation values ​​and the temperature response data of each acquisition point is realized, forming a matching dataset, which provides synchronous data input for curve fitting analysis. To achieve effective comparison between thermal behavior perturbation values ​​and temperature response data, it is first necessary to map the thermal behavior perturbation values ​​and temperature response data to their respective acquisition points using spatial coordinates as a reference. Based on this, a time synchronization index table is constructed by setting a unified time sampling frequency and synchronization timestamp to ensure the correspondence between different data sources in the time dimension. This index table pairs the thermal behavior perturbation values ​​of each acquisition point with the corresponding temperature response data at each time node, forming a matching dataset with a bidirectional correspondence. For example, if the perturbation value at a certain location changes abruptly in the early stages of heating, but the temperature response curve does not show synchronous fluctuations, its response hysteresis characteristic can be identified in subsequent curve fitting. This approach allows subsequent fitting analysis to no longer rely on a single data source, but rather analyzes the temperature curve change trend under the background of thermal perturbation, thereby improving the sensitivity and robustness of the fitting results to thermal anomalies. In the technical features, the thermal behavior perturbation value represents the thermal anomaly risk weight of the collection point, the temperature response data is the measured value of the heating behavior corresponding to that point, the time synchronization index table ensures the temporal consistency between data, the spatial coordinate order is used to maintain the consistent pairing logic of the points, and the matching dataset is the core input structure for subsequent mathematical modeling and judgment correction.

[0041] In the matching dataset, select temperature response data segments within a continuous time window, apply the sliding window method to slide sequentially with a set window width and step size, and use the least squares fitting algorithm to fit the temperature response data with the target heating curve. Record the fitting residual value and response delay difference value at each window position to capture the local heating response shift in time. To accurately capture the offset characteristics of local heating behavior over time, a temperature response data segment within a continuous time window can be selected from the matching dataset, and the sliding window method can be used for analysis. The sliding window method is a local sequence analysis method that, by setting a fixed window width and step size, allows the analysis window to slide progressively along the time axis on the temperature response data, ensuring that each fit focuses on the response characteristics within a local time period. For each data segment in the window, a least-squares fitting algorithm is applied to fit the temperature response data of the sampled points to the target heating curve. The target heating curve is an ideal temperature rise trajectory constructed based on normal heating patterns, with a fixed heating slope and time response model. The least-squares fitting algorithm, by solving for the function that minimizes the sum of squared errors, can assess the degree of deviation between the temperature response data and the target heating curve within the local time window. This process extracts two key parameters: the fitting residual value, which characterizes the numerical deviation between the actual response and the ideal heating in each window; and the response delay difference value, which identifies the hysteresis behavior of the temperature response by comparing the time axis offset between the fitted curve and the target curve. Among the technical features, the sliding window method is used to realize local time series analysis, the least squares fitting algorithm realizes accurate curve comparison modeling, and the target heating curve serves as a standard reference to ensure consistent judgment. The combination of these three constitutes a high-resolution heating offset identification method.

[0042] The fitting residuals and response delay differences of all sliding windows are normalized and calculated to generate thermal response offset curves that reflect the entire heating process of each sampling point. By analyzing the fluctuation frequency and amplitude of the offset curves, stable deviation intervals are extracted and a set of thermal response offset parameters is formed, providing input data basis for subsequent confidence index calculation.

[0043] To extract the thermal response anomaly characteristics of each acquisition point from the overall time dimension, it is necessary to normalize the fitting residuals and response delay differences obtained in all sliding windows. The purpose of normalization is to unify values ​​of different scales and units into the same interval, making them comparable and compatible. Range normalization or Z-score standardization methods are commonly used to standardize the mapping of each indicator. After normalization, the residuals and delay differences of the same acquisition point in multiple time windows can be combined into a continuous data stream to construct a complete thermal response offset curve. This curve reflects the evolution trend of the response offset at the acquisition point throughout the entire heating phase; its fluctuation frequency represents the periodicity of the temperature response anomaly, and the fluctuation amplitude represents the strength of the anomaly response. By extracting frequency features through Fourier transform or sliding standard deviation analysis, and then combining this with local extremum identification to extract fluctuation amplitude, the time intervals where the temperature response offset is relatively stable can be accurately defined. These stable deviation intervals are recorded as a parameter set, and a thermal response offset degree parameter set is constructed by combining time, amplitude, and frequency dimensions, providing structured input for subsequent calculation of the temperature data reliability index. In this process, normalization calculations ensure collaborative analysis of data across all dimensions, thermal response offset curves provide the basis for time series analysis, and deviation interval extraction enables precise definition of abnormal behavior.

[0044] S4. Reconstruct the judgment logic of temperature acquisition based on the reliability index of temperature acquisition, adopt a multi-parameter joint judgment structure to classify and identify the heating state, and adjust the temperature judgment node, heating duration and heat input distribution scheme. In this embodiment, S4 specifically refers to: Based on the reliability index of temperature acquisition, all temperature acquisition points are classified into reliability levels, low reliability points are eliminated and a reliability index table is generated. A one-to-one mapping relationship between the index table and the distribution of thermal behavior disturbance is established, the judgment logic of temperature acquisition is reconstructed, and the reconstructed logic is used as the input premise structure of the judgment system. To effectively utilize the temperature acquisition reliability index, all temperature acquisition points are first classified into levels according to a pre-defined reliability grading rule. The level is based on a comprehensive score of multiple parameters, including the fitting residual between the temperature waveform and the target curve, the degree of thermal response offset, and the weighting factor of local thermal anomaly characteristics. Each acquisition point is then classified into high, medium, or low reliability levels after scoring. Low reliability points are excluded from subsequent logic because their data may be affected by factors such as air resistance interference, sensor drift, or thermal inertia hysteresis, and directly involving them in temperature control judgment would lead to system misjudgment. Next, the remaining medium and high reliability points are numbered and a reliability index table is generated to represent the dual information of each node in terms of spatial coordinates and data reliability. Then, the index table is mapped point-by-point to the previously formed thermal behavior perturbation distribution data, constructing a correspondence between the thermal anomaly area distribution and the reliability distribution. This provides a reliable data foundation and region identification accuracy for the subsequent reconstruction of the judgment logic.

[0045] The "Temperature Acquisition Reliability Index" represents the reliability level of temperature data from each acquisition point during the sulfurization heating process, considering both statistical and dynamic response dimensions. "Reliability Level Classification" refers to classifying and grading acquisition points based on comprehensive feature values. "Eliminating Low-Reliability Points" reflects the strategy of excluding abnormal or interfered points. The "Reliability Index Table" is a structured dataset establishing relationships between the spatial location and reliability level of the remaining acquisition points. "Thermal Behavior Disturbance Distribution" refers to the spatial disturbance zone image calculated based on multidimensional thermal anomaly indicators in the thermal behavior feature field. "One-to-One Mapping Relationship" refers to precisely pairing reliable acquisition points with their positions in the thermal disturbance map. "Reconstructing the Judgment Logic of Temperature Acquisition" means re-establishing the input boundaries and analysis process of the heating determination model after eliminating interference and strengthening the effective data foundation. The "Input Prerequisite Structure of the Judgment System" indicates that this reconstructed logic will serve as the input condition framework for subsequent core decision-making models such as state recognition and control command judgment. Through this technical approach, the data input quality of the control system can be guaranteed from the source, improving the accuracy and robustness of temperature and pressure control judgments.

[0046] Based on the reconstructed judgment logic, the reliability index of temperature acquisition, the degree of thermal response offset and spatial node features are integrated to construct a joint judgment structure with multiple parameter dimensions. Density clustering and support vector classification algorithms are used to classify the heating state and output classification labels for steady state, thermal hysteresis state and nonlinear fluctuation state. To improve the accuracy of temperature state determination during the heating phase, based on the temperature acquisition reliability index, the thermal response offset degree and spatial node features are further integrated in a multi-dimensional manner to construct a joint determination structure containing multiple parameter dimensions. This structure uses each temperature acquisition point as the basic unit to construct a feature vector group. The vector dimensions include thermal response stability index, temperature rise rate offset, thermal diffusion curvature, spatial distribution coordinates, and temperature fluctuation periodicity information. Based on this, a density clustering algorithm is introduced to identify the distribution characteristics of temperature states in a high-dimensional parameter space. Density clustering does not rely on a preset number of categories and is suitable for identifying nonlinear and irregular state boundaries, thus effectively aggregating points with similar thermal behavior patterns. Subsequently, a support vector classification algorithm is used to train labels on these clustering results, constructing a discriminant function for online classification, outputting spatial classification labels for stable states, thermal hysteresis states, and nonlinear fluctuation states, providing a refined identification foundation for the execution of thermal control strategies.

[0047] "Reconstructed Judgment Logic" refers to a new judgment system for temperature state analysis formed after eliminating the credibility of previous data collection points and reorganizing the logical structure. "Temperature Acquisition Credibility Index" reflects the quantitative characteristics of each data point in the dimension of sampling credibility; "Thermal Response Shift" represents the degree of response difference caused by abnormalities in the heat transfer path or local medium structure; "Spatial Node Features" refers to the spatial location, thermal history, and relative layout of each data collection point within the mold and its adjacent nodes; "Joint Judgment Structure" is a model framework that integrates input features from multiple dimensions for unified classification judgment; "Density Clustering Algorithm" is an unsupervised learning method based on local density functions, capable of automatically identifying the natural clustering trend of sample points and adapting to irregular boundaries, with typical algorithms such as DBSCAN; "Support Vector Classification Algorithm" is a supervised learning method that achieves multi-class segmentation based on the idea of ​​maximum margin hyperplane, suitable for high-dimensional small-sample classification tasks; "Heating State Classification Label" is one of the algorithm's output results, corresponding to three typical behavioral states in the temperature control process, facilitating precise formulation of control responses according to the state in subsequent execution stages. This combined strategy enables structured intelligent classification of the heating process at the level of thermal response mechanism, greatly enhancing the sensitivity and reliability of identifying nonlinear heat transfer anomalies.

[0048] The temperature acquisition judgment strategy parameters are updated based on the temperature rise status classification label. By dynamically adjusting the selection strategy and distribution density of temperature judgment nodes, the target range of the temperature rise duration is reset, and the power ratio of each heat input path is adjusted and optimized. This achieves the linkage reconstruction of the heat input distribution scheme in the temporal and spatial dimensions to adapt to the changing characteristics of the actual temperature rise status during the vulcanization process.

[0049] The temperature acquisition strategy parameters are updated based on the temperature rise state classification labels, and dynamic adjustments can be achieved by constructing a feedback-based adaptive thermal control mechanism. First, for spatial regions marked as thermal hysteresis, the distribution density and selection sensitivity of temperature judgment nodes are increased to more densely cover abnormal areas, ensuring real-time capture of temperature fluctuation details. For stable regions, node density is appropriately reduced to decrease data redundancy. Based on this, the target range for the temperature rise duration is reset, specifically by analyzing the changing trends of the thermal response curve and the time delay coefficient, dynamically adjusting the upper and lower limits of the control time for the temperature rise process. Next, the power ratio of the heat input path is adjusted in real time based on the temperature rise state classification results. For example, the heat input power is increased in regions in thermal hysteresis, and a rhythmic heating mode is activated in regions with nonlinear fluctuations to suppress fluctuations, thereby optimizing the thermal energy distribution structure in the spatial dimension. Finally, by integrating the distribution logic of temperature judgment nodes, the temperature rise time control strategy, and the heat input path adjustment into a linked reconstruction model, the thermal control scheme has the ability to be reconstructed in real time according to state changes, fully matching the dynamic temperature rise behavior characteristics during the vulcanization process.

[0050] "Heating State Classification Label" is one of the outputs of the thermal behavior analysis model, indicating the physical state category of the current temperature response, often including steady state, thermal hysteresis state, and nonlinear fluctuation state. "Temperature Acquisition Judgment Strategy Parameters" covers multiple key threshold settings and time control conditions used to determine whether the temperature meets the target, forming the basic rule system of the thermal control logic. "Temperature Judgment Node Selection Strategy and Distribution Density" refers to which points in the mold space are selected as temperature judgment base points and the density of these points; adjusting this parameter controls the judgment granularity. "Target Range of Heating Duration" defines the duration boundary of the thermal process control cycle; dynamic setting allows the heating phase to better match the actual rhythm of the thermal response. "Power Ratio of Heat Input Path" describes the energy distribution ratio of heating modules or heating channels in different areas; optimizing this distribution achieves thermal field balance. "Linked Reconstruction of Heat Input Distribution Scheme in Temporal and Spatial Dimensions" refers to not only adjusting the heating intensity at different locations in the heating control strategy but also starting, stopping, or switching heat sources in a timely manner according to the real-time status, achieving overall coordination and feedback optimization in the time-space dimension. This process enhances the targeting and responsiveness of heat input, effectively avoiding problems such as insufficient local vulcanization or heat accumulation caused by static strategies.

[0051] S5. Calculate the temperature and pressure balance coefficient based on the judgment logic correction result, and compare the temperature output and pressure response synchronously during the heating stage to execute a real-time temperature and pressure adjustment strategy to achieve dynamic control of the heating process.

[0052] In this embodiment, S5 specifically refers to: Based on the temperature acquisition reliability index and thermal response offset of each spatial node in the judgment logic correction result, the temperature change gradient and pressure response gradient in the time series are jointly extracted to construct a set of temperature and pressure characteristic parameters. The temperature and pressure balance coefficient is calculated by weighted fusion method to quantify the degree of change in the coupling relationship between temperature output and pressure response during the heating stage. During the vulcanization heating stage of Ebonite rubber rollers, the synergistic relationship between temperature output and pressure response directly affects the effective heat transfer and the stability of the molding process. To accurately assess this relationship, the reliability index of temperature acquisition and the degree of thermal response offset of each spatial node can be obtained based on the judgment logic correction results. Combined with real-time acquired time-series temperature and pressure data, the temperature change gradient and pressure response gradient of each node within the same time period are extracted. The temperature change gradient can be calculated using the temperature difference and time difference between adjacent sampling points, while the pressure response gradient is determined by monitoring the rate of change of pressure loading. After normalizing these parameters, they are fused according to a set weighting coefficient to generate a comprehensive temperature-pressure balance coefficient. This coefficient quantifies the coordination between the current heat input and pressure response; a high balance value indicates good heat-pressure synchronization, while a low balance value indicates a mismatch between heat conduction and pressure loading. For example, in the initial stage of heating, if the pressure rises rapidly while the temperature response lags behind, the balance coefficient will decrease, indicating that the control system needs to be adjusted to avoid insufficient local vulcanization.

[0053] The temperature acquisition reliability index is a quantitative indicator reflecting the reliability of temperature data at each acquisition point. It is derived from the fitting residual and response delay calculation results between the temperature response curve and the target heating curve. The thermal response offset measures the degree of deviation of the actual temperature response from the expected state. Together, they reflect the reliability and offset risk of the acquired data. The temperature change gradient represents the rate of temperature change per unit time and is an important basis for evaluating the heating rate and thermal diffusion trend. The pressure response gradient quantifies the temporal dynamic characteristics of pressure change, reflecting the changing state of external pressure application. After classifying these parameters into a temperature and pressure characteristic parameter set, different weights are assigned to each parameter dimension to form a fused temperature and pressure balance coefficient. This coefficient not only reflects the change of a single data point but also reflects the interactive coupling characteristics between multiple parameters, providing a precise input basis for thermal and pressure matching control during the heating process.

[0054] During the heating phase, the temperature output and pressure response are synchronously compared based on the temperature-pressure balance coefficient. By establishing a synchronization time index table, the dynamic change trajectories of temperature and pressure at each moment are compared. The sliding window algorithm is used to identify the delay alignment point and synchronization offset trend of the temperature-pressure response curve, and generate a temperature-pressure alignment offset curve for judging the thermal-pressure coordination during the heating process. During the heating phase of the Ebonite rubber roller vulcanization process, to achieve effective coupling control of temperature output and pressure response, it is necessary to synchronize and compare their dynamic trajectories in real time. This process can be achieved by constructing a synchronization time index table, pairing temperature acquisition time points with corresponding pressure response time points one by one, forming a complete time synchronization data structure. Based on this, a sliding window algorithm is applied to segment and analyze the temperature and pressure response curves. The time delay difference between the two curves is compared sequentially within the window, identifying the temperature-pressure delay alignment point and synchronization offset trend within each time period. For example, if the pressure response curve significantly lags behind the temperature curve during a certain period in the early stages of vulcanization, this period can be marked as a temperature-pressure mismatch interval through window detection. The synchronization offset trends extracted within these time periods are normalized and connected on the time axis to form a continuous temperature-pressure alignment offset curve, dynamically reflecting the real-time coupling coordination between heat and pressure. The smaller the fluctuation of this curve, the more synchronized the temperature and pressure responses; the larger the fluctuation, the more incompatible and abnormal behavior exists between heat and pressure, requiring the triggering of subsequent control strategies.

[0055] The temperature-pressure balance coefficient is a composite index calculated based on the dynamic characteristics of thermal behavior and pressure changes, used to measure the stability of the thermo-pressure matching state. The synchronization time index table is a data structure established by establishing a one-to-one correspondence between temperature output and pressure response under the same time reference, making the data from different acquisition channels comparable in time sequence. Temperature output and pressure response are represented as two continuous time series signals. The sliding window algorithm divides the two curves into several segments along the time dimension by setting a fixed width and sliding step size, analyzing their time differences and morphological matching in segment by segment, which can accurately capture instantaneous coupling mismatch points. The delay alignment point represents the optimal synchronization overlap position of temperature and pressure on the time axis, used to determine whether the response time difference is within the tolerable range; the synchronization offset trend records the trend of synchronization changes within each window segment on the time axis, used as a basis for subsequent adjustment. The final output temperature-pressure alignment offset curve reflects the evolution trajectory of thermo-pressure dynamic coordination throughout the heating process, and is an important basic data for realizing closed-loop regulation of process control.

[0056] Based on the synchronous offset values ​​of each time period in the temperature and pressure alignment offset curve, an adaptive adjustment algorithm is used to construct a real-time temperature and pressure adjustment strategy. The heat input power distribution and pressure loading rate during the heating stage are dynamically adjusted according to the sampling period. When the temperature and pressure coupling offset exceeds the control tolerance range, the control parameters are corrected in time to achieve synchronous temperature and pressure control and dynamic stable matching during the heating process.

[0057] To achieve coordinated control of heat input and pressure loading during the heating phase, a real-time temperature and pressure adjustment strategy with responsiveness and self-adjustment capabilities needs to be constructed based on the synchronization offset values ​​of each time period in the temperature-pressure alignment offset curve. This strategy can be implemented through an adaptive adjustment algorithm. First, the temperature-pressure alignment offset value for the current time period is read in each sampling period and compared with a preset tolerance threshold. If the offset value is within the tolerance range, the current heat input power distribution and pressure loading rate are maintained; if the offset value exceeds the threshold, the spatial distribution ratio of heat input power is adjusted in real time according to the offset direction and offset magnitude. For example, the power density is reduced in the overheated region, and the pressure loading rate is increased in the stagnant region, thereby pushing the temperature and pressure response trajectories back towards synchronization. For example, if a rapid temperature rise and a lagging pressure response are observed in a certain time period, the system can automatically reduce the output power of the heating unit in that segment and fine-tune the loading rhythm of the pressure head to restore the timing consistency between the two. This mechanism dynamically corrects the control parameters in each sampling period through closed-loop control, ensuring a dynamically stable matching state between temperature and pressure responses and preventing local overheating or pressure imbalance.

[0058] The temperature-pressure alignment offset curve is a time-series curve constructed by comparing the temperature and pressure responses over time using a sliding window algorithm, reflecting the synchronicity of thermo-pressure coupling. The synchronization offset value is the response offset between temperature and pressure calculated in each time period, including indicators such as time delay and response difference. The adaptive adjustment algorithm is a control model based on the dynamic changes of offset trends and control targets. It can automatically calculate the adjustment proportional coefficient according to the actual deviation amplitude, realizing real-time correction of control parameters. The heat input power distribution refers to the proportion of electric heating or heat flow input distributed among different heating units or spatial regions, while the pressure loading rate represents the rate of change of pressure applied per unit time. The control tolerance range is a preset allowable fluctuation range for temperature-pressure synchronicity; exceeding this range is considered a mismatch state and requires triggering a correction mechanism. Through continuous feedback and adjustment within the sampling period, the coupling balance between heat supply and pressure changes during the heating process is achieved, thereby improving the stability and accuracy of the vulcanization molding process.

[0059] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions according to the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means (e.g., infrared, wireless, microwave, etc.). A computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. Available media can be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media. Semiconductor media can be solid-state drives.

[0060] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0061] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0062] In the several embodiments provided in this application, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0063] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0064] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0065] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for producing Ebonite rubber rollers based on intelligent temperature and pressure control, characterized in that, Specifically, the following steps are included: S1. Collect time-series temperature data and pressure response data at different spatial positions between the Ebonite rubber roller and the mold, and construct a thermal behavior feature field containing temperature gradient change characteristics and pressure phase characteristics to determine whether there is gas resistance causing abnormal heat conduction during the vulcanization heating process. S2. In the case of abnormal heat conduction caused by air resistance, a difference mapping model is established based on the thermal behavior characteristic field to calculate the temperature waveform delay, thermal diffusion curvature and temperature rise slope offset, and generate a local temperature anomalous distribution matrix to determine the local temperature anomalous changes in the case of abnormal heat conduction caused by air resistance during the sulfurization heating process. S3. Construct a temperature acquisition reliability function based on the local temperature anomalous distribution matrix, fit and compare the temperature response of each acquisition point with the target heating curve, generate a reliability index for judging the reliability of temperature data, and use it to correct the judgment logic of temperature acquisition based on the local temperature anomalous changes caused by gas resistance during the sulfurization heating process. S4. Reconstruct the judgment logic of temperature acquisition based on the reliability index of temperature acquisition, adopt a multi-parameter joint judgment structure to classify and identify the heating state, and adjust the temperature judgment node, heating duration and heat input distribution scheme. S5. Calculate the temperature and pressure balance coefficient based on the judgment logic correction result, and compare the temperature output and pressure response synchronously during the heating stage to execute a real-time temperature and pressure adjustment strategy to achieve dynamic control of the heating process.

2. The method for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to claim 1, characterized in that, S1 specifically refers to: Temperature and pressure acquisition units are arranged radially and axially in the contact area between the Ebonite rubber roller and the mold. Temperature and pressure response data at each spatial location are acquired synchronously by setting a time series sampling period. The correspondence between sampling points and spatial coordinates is established in time order to form a time series temperature and pressure response dataset for different spatial locations. The time-series temperature data and pressure response data are matched and calculated according to spatial coordinates. By calculating the rate of change of temperature gradient and pressure phase difference between adjacent acquisition units, a multi-dimensional parameter matrix containing temperature gradient change characteristics and pressure phase characteristics is constructed. This multi-dimensional parameter matrix is ​​then spatially mapped to form a thermal behavior feature field, which is used to reflect the thermal and pressure coupling relationship between the Ebonite rubber roller and the mold. By jointly comparing and analyzing the temperature gradient change characteristic parameters and pressure phase characteristic parameters in the thermal behavior characteristic field, when a discontinuous abrupt change in the temperature gradient change characteristic and a synchronous shift in the pressure phase characteristic are detected, it is determined that there is an abnormal situation of heat conduction caused by gas resistance during the sulfurization heating process.

3. The method for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to claim 1, characterized in that, S2 specifically includes the following steps: S201. In the case of abnormal heat conduction caused by air resistance, a difference mapping model is constructed based on the thermal behavior feature field. By comparing the temperature gradient change characteristics and pressure phase characteristics of the normal and abnormal regions in the thermal behavior feature field under the same spatial path, the spatial response difference parameter is extracted and mapped to a unified thermal behavior reference grid to reflect the heat conduction path change characteristics under air resistance interference. S202. Based on the difference mapping model, the time series offset analysis method is used to analyze the temperature change curve of each spatial node in the thermal behavior feature field, calculate the temperature waveform delay to identify the heat conduction time lag, calculate the heat diffusion curvature based on the three-point method to identify the heat flow return or focusing area, and obtain the temperature rise slope offset through differential calculation to quantify the local heating abnormal trend. S203. The temperature waveform delay, thermal diffusion curvature and temperature rise slope offset are constructed into a local temperature anomalous distribution matrix according to the spatial node order. By matching the results of multi-index coupling in the matrix with the threshold, the local temperature anomalous change region under the condition of abnormal heat conduction caused by gas resistance during the sulfurization heating process is identified, and the spatial coordinates and thermal feature labels of the corresponding region are output.

4. The method for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to claim 3, characterized in that, S202 specifically refers to: Based on the difference mapping model, the temperature change curve of each spatial node in the thermal behavior feature field is time normalized. By setting a uniform sampling time interval, the temperature change sequences of different nodes are aligned with time index to form a temperature sequence set with a synchronous time reference for subsequent time series offset analysis. In the temperature sequence set, temperature change curve pairs of adjacent spatial nodes are selected. The time offset between each curve is calculated by sliding window to obtain the temperature waveform delay parameter. The numerical change rate of adjacent temperature points is fitted by the three-point method to calculate the thermal diffusion curvature, thereby obtaining the local curvature distribution of the heat flow direction. Differential operation is performed on the time-continuous sampled values ​​of each node in the temperature sequence set to obtain the change in temperature rise rate and couple it with the thermal diffusion curvature data to extract the temperature rise slope offset to characterize the local heating trend change features, and output it in the form of a parameter group for subsequent matrix generation process.

5. The method for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to claim 1, characterized in that, S3 specifically includes the following steps: S301. Based on the temperature waveform delay, thermal diffusion curvature and temperature rise slope offset of each spatial node in the local temperature anomaly distribution matrix, construct a three-dimensional thermal anomaly feature vector set, set the confidence weights corresponding to each feature dimension, and generate node-level thermal behavior perturbation values ​​through vector weighting calculation, which are used to construct the temperature acquisition confidence function. S302. Compare the thermal behavior perturbation value output by the temperature acquisition confidence function with the temperature response data of each acquisition point, and perform curve fitting between the temperature response data and the target heating curve based on the sliding window method. Calculate the fitting residual and response delay difference index to quantify the degree of thermal response offset of the acquisition point. S303. The thermal response offset degree and the temperature acquisition credibility function result are weighted and fused to generate a credibility index for judging the credibility of temperature data. The judgment logic of temperature acquisition is corrected according to the credibility index, and the acceptance level of acquisition points, judgment trigger conditions and time fault tolerance range are adjusted to realize the responsive identification of local abnormal temperature changes and the optimization of acquisition strategy.

6. The method for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to claim 5, characterized in that, S302 specifically refers to: The thermal behavior perturbation values ​​output by the temperature acquisition confidence function are paired and compared with the temperature response data corresponding to each acquisition point in spatial coordinate order. By establishing a time synchronization index table, a one-to-one correspondence between the thermal behavior perturbation values ​​and the temperature response data of each acquisition point is realized, forming a matching dataset, which provides synchronous data input for curve fitting analysis. In the matching dataset, select temperature response data segments within a continuous time window, apply the sliding window method to slide sequentially with a set window width and step size, and use the least squares fitting algorithm to fit the temperature response data with the target heating curve. Record the fitting residual value and response delay difference value at each window position to capture the local heating response shift in time. The fitting residuals and response delay differences of all sliding windows are normalized and calculated to generate thermal response offset curves that reflect the entire heating process of each sampling point. By analyzing the fluctuation frequency and amplitude of the offset curves, stable deviation intervals are extracted and a set of thermal response offset parameters is formed, providing input data basis for subsequent confidence index calculation.

7. The method for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to claim 1, characterized in that, S4 specifically refers to: Based on the reliability index of temperature acquisition, all temperature acquisition points are classified into reliability levels, low reliability points are eliminated and a reliability index table is generated. A one-to-one mapping relationship between the index table and the distribution of thermal behavior disturbance is established, the judgment logic of temperature acquisition is reconstructed, and the reconstructed logic is used as the input premise structure of the judgment system. Based on the reconstructed judgment logic, the reliability index of temperature acquisition, the degree of thermal response offset and spatial node features are integrated to construct a joint judgment structure with multiple parameter dimensions. Density clustering and support vector classification algorithms are used to classify the heating state and output classification labels for steady state, thermal hysteresis state and nonlinear fluctuation state. The temperature acquisition judgment strategy parameters are updated based on the temperature rise status classification label. By dynamically adjusting the selection strategy and distribution density of temperature judgment nodes, the target range of the temperature rise duration is reset, and the power ratio of each heat input path is adjusted and optimized. This achieves the linkage reconstruction of the heat input distribution scheme in the temporal and spatial dimensions to adapt to the changing characteristics of the actual temperature rise status during the vulcanization process.

8. The method for producing Ebonite rubber rollers based on intelligent temperature and pressure control according to claim 1, characterized in that, S5 specifically refers to: Based on the temperature acquisition reliability index and thermal response offset of each spatial node in the judgment logic correction result, the temperature change gradient and pressure response gradient in the time series are jointly extracted to construct a set of temperature and pressure characteristic parameters. The temperature and pressure balance coefficient is calculated by weighted fusion method to quantify the degree of change in the coupling relationship between temperature output and pressure response during the heating stage. During the heating phase, the temperature output and pressure response are synchronously compared based on the temperature-pressure balance coefficient. By establishing a synchronization time index table, the dynamic change trajectories of temperature and pressure at each moment are compared. The sliding window algorithm is used to identify the delay alignment point and synchronization offset trend of the temperature-pressure response curve, and generate a temperature-pressure alignment offset curve for judging the thermal-pressure coordination during the heating process. Based on the synchronous offset values ​​of each time period in the temperature and pressure alignment offset curve, an adaptive adjustment algorithm is used to construct a real-time temperature and pressure adjustment strategy. The heat input power distribution and pressure loading rate during the heating stage are dynamically adjusted according to the sampling period. When the temperature and pressure coupling offset exceeds the control tolerance range, the control parameters are corrected in time to achieve synchronous temperature and pressure control and dynamic stable matching during the heating process.

Images