Cloud data driven twin construction method and system for insulation board production line

Abstract

The present application relates to the technical field of industrial production digitization, and particularly relates to a method and system for twin construction of a heat insulation board production line based on cloud data driving. The method comprises the following steps: acquiring continuous slab flow data; performing sensing section binding on the continuous slab flow data to obtain sensing section data; performing multi-dimensional time calendar footprint reconstruction according to the sensing section data to obtain time calendar footprint data; performing cross-process heterogeneous coupling simulation on the time calendar footprint data to obtain cross-process simulation data; performing global heterogeneous data field reconstruction according to the cross-process simulation data to obtain continuous slab overall field data; and performing twin continuous expression according to the continuous slab overall field data to obtain slab twin data. The present application realizes continuous tracking and unified expression of multi-dimensional states in the production process by constructing a continuous slab twin based on sectional modeling and cross-process coupling simulation, and improves the modeling accuracy of the production line.

Description

Technical Field

[0001] This invention relates to the field of digital technology in industrial production, and in particular to a method and system for constructing a twin of an insulation board production line based on cloud data. Background Technology

[0002] The production process of continuous slab products such as insulation boards typically involves multiple continuous processes, including laying, pressing, curing, and conveying. These processes exhibit clear temporal dependencies and physical coupling. Current technologies for digital management of such production lines often employ data acquisition and monitoring methods based on single equipment or single processes, focusing on discrete time points or local equipment states. This approach fails to provide a unified description of the overall evolution of the continuous slab across both spatial and temporal dimensions.

[0003] With the development of digital twin technology, some solutions attempt to map and simulate production processes by constructing virtual models. However, existing digital twin methods are mostly designed for discrete manufacturing objects, typically using the state of a single product or equipment as the modeling unit. For non-discrete objects like continuous slabs that evolve continuously during transport and whose boundaries change dynamically, there is a lack of effective modeling methods. Furthermore, in situations where multi-source sensing data suffers from inconsistent sampling frequencies, difficulties in time alignment, and data fragmentation, it is challenging to achieve accurate tracking and reconstruction of continuous states. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a cloud data-driven method and system for constructing a twin of an insulation board production line, thereby resolving at least one of the aforementioned technical issues.

[0005] This application provides a cloud data-driven method for constructing a twin of an insulation board production line, including the following steps: Step S1: Obtain continuous slab flow data; bind sensing sections to the continuous slab flow data to obtain sensing section data; Step S2: Reconstruct the multidimensional time-history footprint based on the sensing segment data to obtain the time-history footprint data; Step S3: Perform cross-process heterogeneous coupling simulation on the time-history footprint data to obtain cross-process simulation data; reconstruct the global heterogeneous data field based on the cross-process simulation data to obtain the overall field data of the continuous slab. Step S4: Perform continuous twin representation based on the overall field data of the continuous slab to obtain slab twin data.

[0006] This invention transforms the previously difficult-to-model continuous medium production process into a segment-level dynamic evolution system with traceable and computable characteristics by segmenting and binding continuous slab flow data, reconstructing multi-dimensional time-history footprints, and performing cross-process heterogeneous coupling simulation. This enables continuous representation of the slab's state changes in different processes. Based on a global heterogeneous data field reconstruction mechanism, multi-dimensional states such as thickness, temperature, and density are fused and expressed within a unified spatial framework. Overall continuity is maintained through boundary fading and cross-process transmission, effectively avoiding the boundary abrupt changes and information fragmentation problems caused by traditional discrete modeling. The constructed twin can evolve in real time with the production process, improving not only the accuracy of the process chain representation but also enhancing the ability to represent anomaly propagation paths and quality formation mechanisms.

[0007] Optionally, step S1 specifically includes: Acquire continuous slab flow data; Heterogeneous sensing time-domain normalization is performed on continuous slab flow data to obtain heterogeneous sensing time-domain data. Virtual segments are generated from heterogeneous sensing time-domain data to obtain virtual segment data. Based on the virtual segment data, virtual and real space are anchored to obtain the perceived segment data.

[0008] This invention achieves unified temporal alignment and structured representation of multi-source asynchronously acquired data by performing heterogeneous sensing temporal normalization, virtual segment generation, and virtual-real spatial anchoring on continuous slab flow data. This transforms the originally continuous and difficult-to-identify slab flow process into segment units with clear boundaries and unique identifiers. This process not only effectively eliminates data misalignment problems caused by differences in sampling frequencies and time offsets between different sensor sources, but also enhances the discrete representation capability of continuous media through the construction of virtual segments, enabling analysis to be performed at a uniform granularity. Virtual-real spatial anchoring establishes a stable mapping relationship between segment data and actual production line spatial locations, improving the traceability and consistency of segments during cross-process flow and enhancing the accuracy and stability of overall twin modeling.

[0009] Optionally, step S2 specifically includes: Based on the sensing segment data, segment anchor points are extracted to obtain segment anchor point data; Multimodal inertial fingerprint encoding is performed on the segment anchor point data to obtain segment inertial fingerprint data; Segment continuation probability is deduced based on segment inertial fingerprint data to obtain segment continuation relationship data; Based on the segment continuity data, cross-process fractured segments are sutured to obtain segment suture data; Multi-scale state history unfolding was performed on the segment suture data to obtain time-history footprint data.

[0010] This invention employs segment anchor point extraction and multimodal inertial fingerprint encoding to stably extract and structurally represent key state features of continuous slabs during time progression. This allows segments to be identified without relying on single-location or instantaneous data, instead possessing the ability to maintain consistent state identification across time windows. Dynamic relationships between segments are constructed through segment continuity probability deduction, and combined with cross-process fracture segment stitching, trajectory interruptions caused by equipment switching, sampling blind spots, or process disturbances are reconstructed and repaired, thereby restoring the continuous evolution path of segments within the complete production chain. Through multi-scale state history unfolding, the segment-level evolution process is elevated to a holistic expression level across time and processes, improving not only the traceability and consistency of continuous slabs in complex process environments but also enhancing the ability to express state evolution trends and anomaly propagation processes.

[0011] Optionally, the segment continuation probability deduction is as follows: Segment inertial kernels are extracted from the segment inertial fingerprint data to obtain segment inertial kernel data; The forward reachability domain is expanded based on the segment inertial kernel data to obtain the forward reachability domain data; Multi-channel preservation of forward reachable domain data is calculated to obtain continuity preservation data. Based on the continuity retention data, inertial decay compensation is extrapolated to obtain compensation retention data; Based on the compensation preservation data, neighboring segment cooperative exclusion inference is performed to obtain neighboring segment competition relationship data; Based on the data on the competition relationship between adjacent segments, discontinuous crossing and bridging simulation is performed to obtain crossing and bridging data; Based on the bridging data, the continuity confidence is aggregated to obtain the segment continuity probability data; The main continuation chain is extracted based on the segment continuation probability data to obtain segment continuation relationship data.

[0012] In this invention, by extracting the inertial kernel and expanding the forward reachable domain of the segment's inertial fingerprint, this step extends the possible evolution range of the segment over time from a single location to a candidate space with tolerance and deformation adaptability, effectively adapting to the non-rigid change characteristics of continuous slabs under pressing, traction, and thermal effects. Combining multi-channel retention calculation and inertial attenuation compensation simulation, the comparability of different physical states remains under cross-time windows and cross-process conditions, thus avoiding misjudging normal process-induced changes as segment fractures. By constraining the sequential consistency between segments through neighboring segment cooperative exclusion, and introducing discontinuous bridging simulation to compensate for sampling blind spots or short-term interruptions, continuous recovery of the segment evolution chain is achieved. Through continuation credibility aggregation and main continuation chain extraction, a stable and globally consistent segment continuation relationship is constructed, which not only improves the accuracy and robustness of continuous slab tracking but also enhances the expressive ability of the segment evolution logic under working conditions.

[0013] Optionally, the cross-process fracture suturing specifically involves: Breakpoint pairing is performed based on segment continuity data to obtain breakpoint pairing data; The process traversal interpretable domain is constructed from the breakpoint pairing data to obtain joint interpretable domain data; Based on the joint interpretation domain data, a two-way evolutionary docking simulation is performed to obtain docking simulation data; Based on the docking simulation data, implicit bridging segments in the interruption zone are generated to obtain bridging segment data; Based on the bridging segment data, neighboring segment collaborative closure verification is performed to obtain segment stitching data.

[0014] This invention elevates the originally simple problem of time or location interruption to a fracture identification and reconnection process based on process mechanism constraints by constructing an interpretable domain through breakpoint pairing and process traversal. This transforms fragment matching from relying on a single similarity score into one constrained by the rationality of cross-process state evolution. Through bidirectional evolutionary docking and deduction, the segment state is extrapolated and traced back from both before and after the fracture, forming a verifiable continuous transition relationship within the interrupted interval, effectively avoiding mismatch problems caused by unidirectional extrapolation. The system constructs implicit bridging fragments to structurally compensate for data loss caused by sampling blind spots or equipment switching, ensuring that the fracture interval has an interpretable intermediate evolution path. Through neighboring segment collaborative closure verification, the stitching result is constrained from the perspective of the continuity of the overall slab flow, ensuring that local splicing does not disrupt the global order and state consistency, thereby improving the accuracy and stability of cross-process segment connections.

[0015] Optionally, cross-process heterogeneous coupling simulation specifically includes: The process location of each segment is determined from the time-travel footprint data to obtain the segment process data; The segment state transition model and continuous coupling constraint data are constructed by performing segment process data; Based on the state transition model and continuous coupling constraint data, cross-process state recursion calculations are performed to obtain cross-process simulation data.

[0016] This invention locates the process flow within a segment and constructs a segment-level state transition model, explicitly expressing the physical changes under different processes as recursive state evolution relationships. This avoids the information fragmentation problem caused by traditional methods that rely solely on single-point data or single-process analysis. Simultaneously, through continuous coupling constraints, the changes in thickness, temperature, and density of adjacent segments are linked and constrained, ensuring that the simulation results not only satisfy local process rules but also maintain the spatial consistency of the overall continuous slab. By performing cross-process state recursive calculations, the state changes of each process are systematically connected, achieving unified modeling from local processes to the entire process. This enhances the ability to represent state transmission relationships in the process chain, thereby improving the realism and stability of the simulation results.

[0017] Optionally, the reconstruction of the entire heterogeneous data field specifically includes: Based on cross-process simulation data, segment state field elements are extracted to obtain segment field data. Based on the segment field metadata, heterogeneous state variables are mapped to the same domain to obtain the same domain field metadata; Based on the same domain field metadata, the segment boundaries are gradually merged to obtain boundary fusion data; Based on the boundary fusion data, cross-process layered field splicing is performed to obtain the overall field data of the continuous slab.

[0018] This invention employs segmental state field element extraction and heterogeneous state quantity co-domain mapping. This step unifies data from different processes, dimensions, and expression forms into a single space and semantic framework, achieving collaborative expression of multi-dimensional states such as thickness, temperature, and density. This avoids the information fragmentation problem caused by the isolation of traditional multi-source data. Through a segment boundary fading fusion mechanism, boundary abrupt changes occurring during the splicing of adjacent segments are smoothly processed, effectively eliminating state jumps caused by the splicing of discrete segments, thus maintaining the spatial consistency of the continuous slab. By splicing layered fields across processes, the state evolution results of each process stage are systematically integrated, enabling the overall field to not only reflect the spatial distribution at a single moment but also the evolutionary relationships within the process chain.

[0019] Optionally, the gradual fading and merging of segment boundaries specifically refers to: Boundary fusion bands are constructed from the metadata of the same domain field to obtain boundary fusion band data; The difference between the two sides of the boundary is calculated based on the boundary fusion zone data to obtain the difference data between the two sides of the boundary; Distance difference transition data is obtained by calculating the distance difference transition based on the difference data on both sides of the boundary. Boundary fusion data is obtained by performing bilateral coupling fusion calculations based on distance difference transition data.

[0020] This invention constructs a boundary fusion zone and quantifies the state differences on both sides of the boundary. This step transforms the segment splicing problem from a simple boundary splicing to a continuous transition modeling process based on difference distribution, effectively identifying the variation characteristics of multi-dimensional states such as thickness, temperature, and density at the boundary. By introducing joint weights of spatial location and state differences through distance difference transition calculation, different segments can achieve adaptive transition within the fusion zone according to distance and degree of difference, thereby avoiding the over-smoothing or local distortion problems caused by traditional linear interpolation. Through bilateral coupled fusion calculation, the states of the segments on both sides are collaboratively integrated, so that the boundary region maintains both overall continuity and preserves local variation trends.

[0021] Optionally, step S4 specifically includes: The slab mapping data is obtained by meshing the overall field data of the continuous slab. Twin structure data is obtained by constructing a twin structure from the slab mapping data. The twin structure data is recursively updated to obtain the twin update data; Continuous field reconstruction is performed based on the twin update data to obtain slab twin data.

[0022] This invention transforms the originally continuously distributed multidimensional state into a discrete node system with clear spatial topological relationships by performing gridded mapping and structured representation of the overall field data of a continuous slab. This allows for a unified representation of the geometric shape and state distribution of the slab in twin space. A twin structure is constructed, and the node states are continuously evolved through recursive updates, enabling the twin to dynamically represent the changes in the slab at different process stages in the actual production line. Through continuous field reconstruction, the discrete node states are remapped to a continuous distribution, maintaining the consistency of the overall field and enhancing the ability to express local detail changes.

[0023] Optionally, this application also provides a cloud data-driven twin construction system for insulation board production lines, used to execute the cloud data-driven twin construction method for insulation board production lines as described above. The cloud data-driven twin construction system for insulation board production lines includes: At least one processor; A memory, wherein the memory stores computer programs that can be executed by the processor; Communication interface, used for data interaction with production line sensing devices and cloud data platform; When the processor executes the computer program, it performs the following functions: The sensing section binding module is used to acquire continuous slab flow data; and to bind sensing sections to the continuous slab flow data to obtain sensing section data. The multidimensional time-history footprint reconstruction module is used to reconstruct the multidimensional time-history footprint based on the sensing segment data to obtain the time-history footprint data. The cross-process heterogeneous coupling simulation module is used to perform cross-process heterogeneous coupling simulation on the time-history footprint data to obtain cross-process simulation data; and to reconstruct the global heterogeneous data field based on the cross-process simulation data to obtain the overall field data of the continuous slab. The continuous twin expression module is used to continuously express twins based on the overall field data of the continuous slab, and obtain slab twin data.

[0024] The beneficial effects of this invention are as follows: In step S1, by binding sensing segments, continuous data from multiple sources that are heterogeneous and have inconsistent time sequences are uniformly mapped into segment units with stable spatial anchoring relationships, solving the problem of difficulty in identifying and tracking continuous media; In step S2, by reconstructing multi-dimensional historical footprints, the evolutionary relationship of segments in the time and process dimensions is structurally unfolded, and the complete evolutionary link is restored by combining continuation extrapolation and fracture stitching mechanisms, enabling segments to have continuous traceability across time and processes. In step S3, through cross-process heterogeneous coupling simulation, the state transition relationship under different process stages and the continuity constraints between segments are jointly modeled, and through global heterogeneous data field reconstruction, the continuous expression of multi-dimensional states such as thickness, temperature, and density in a unified spatial framework is realized, effectively avoiding the boundary abruptness problem caused by traditional discrete splicing. In step S4, a twin that can dynamically evolve with the production process is constructed through gridded mapping and recursive updates, and a virtual expression result that is highly consistent with the actual slab state is formed through continuous field reconstruction. Attached Figure Description

[0025] Other features, objects, and advantages of this application will become more apparent from the following detailed description of the non-limiting embodiments, taken with reference to the accompanying drawings: Figure 1 A flowchart illustrating the steps of a cloud data-driven twin construction method for an insulation board production line is shown in one embodiment. Figure 2 A flowchart illustrating the steps of a sensing segment binding method according to an embodiment is shown. Figure 3 A flowchart illustrating the steps of a multidimensional time-history footprint reconstruction method according to one embodiment is shown. Figure 4 A flowchart illustrating the steps of a cross-process heterogeneous coupling simulation method according to one embodiment is shown. Figure 5 A flowchart illustrating the steps of a twin continuous expression method according to one embodiment is shown. Detailed Implementation

[0026] The technical method of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. Based on the embodiments of this invention, all other embodiments obtained by those skilled in the art without inventive effort are within the scope of protection of this invention.

[0027] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0028] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0029] Please see Figures 1 to 5 This application provides a cloud data-driven method for constructing a twin of an insulation board production line, including the following steps: Step S1: Obtain continuous slab flow data; bind sensing sections to the continuous slab flow data to obtain sensing section data; In one embodiment, the system acquires time-series data of the conveyor belt's operating speed and corresponding multi-source sensor information, including slab thickness, temperature, and visual features, based on a cloud platform. The visual features originate from visual image data pre-loaded by cameras on the production line. These features are extracted after preprocessing the original images (e.g., grayscale conversion, noise reduction, edge extraction, or texture encoding), and include features such as edge strength, texture distribution, or defect identification. All types of data are resampled and time-aligned using a unified time axis. The time series is integrated based on the conveyor belt speed to establish a mapping relationship between time and spatial location, thereby converting the original time-domain data / time-series data into spatial location-domain data. Based on this spatial mapping result, the continuous slab is divided into segments according to a preset time window or displacement window, forming multiple continuous segments, and each segment is assigned a unique identifier. Various types of sensor data are mapped to their corresponding segments according to their spatial location. For data spanning multiple segments, allocation is based on their proportion within each segment. A segment data structure containing multi-source information such as thickness, temperature, and visual features is constructed for each segment.

[0030] Step S2: Reconstruct the multidimensional time-history footprint based on the sensing segment data to obtain the time-history footprint data; In one embodiment, the system reconstructs the evolution process of a continuous slab based on data from sensing segments. Anchor point features characterizing the state of each segment are extracted, including the average thickness, spatial temperature variation trends, and surface texture features. Simultaneously, within a sliding time window, the thickness variation amplitude, temperature variation amplitude, and state fluctuation degree of each segment are calculated (obtained by calculating the discreteness of the segment's state sequence within the time window, such as calculating the standard deviation or range), thus obtaining the segment's inertial characteristics. The system calculates the spatial range to be reached by the next segment based on the conveyor belt's operating status (the conveyor belt's operating status is acquired by the production line encoder or speed sensor, and the operating status includes at least the real-time speed of the conveyor belt at any given time and its changing trend). Segments are selected within this range, and the continuity between segments is evaluated by comparing retention indicators such as thickness and temperature differences. For example, the system calculates differences between candidate segments and the current segment across multiple state dimensions, including thickness difference, temperature difference, and texture feature differences. When all of these differences are less than preset thresholds (e.g., thickness difference less than 3%, temperature difference less than 10℃, texture change not exceeding 5% or 10%), the two segments are considered continuous. If only partially met, the state change trend is used for auxiliary judgment; for example, consistent thickness change direction can also be considered weak continuity. For feature offsets caused by process changes, corresponding compensation rules are used for correction. For example, thickness reduction is allowed in the pressing process (e.g., a reduction of up to 10%), and temperature is allowed to gradually decrease over time in the curing process. When calculating continuity, the system first performs process compensation correction on the candidate segment's state before participating in the difference calculation. When a break occurs between segments, the system searches for segments that meet continuity conditions within adjacent time windows. (If no segment meets the continuity conditions within the current time window, the system expands the search to one or two subsequent time windows, searching for segments within the corresponding spatially accessible range and filtering them according to the same continuity criteria. If a segment meets both positional and state continuity conditions within the expanded window, a bridging association across time windows is established.) The segments are then connected sequentially according to time to obtain the segment paths representing the evolution of the slab's state over time, i.e., chronological footprint data.

[0031] Step S3: Perform cross-process heterogeneous coupling simulation on the time-history footprint data to obtain cross-process simulation data; reconstruct the global heterogeneous data field based on the cross-process simulation data to obtain the overall field data of the continuous slab. In one embodiment, the system determines the specific process of each segment based on time-travel footprint data, such as the stages of paving, pressing, and curing, and establishes state update rules for different processes. Determining the specific process of each segment involves matching the segment's location with a pre-established production line process space division table based on the segment's spatial location and time sequence information recorded in the time-travel footprint data. The process space division table includes at least the spatial ranges of the paving area, pressing area, and curing area (e.g., 0-10m for the paving area, 10-20m for the pressing area, and 20-40m for the curing area). When the spatial location of a segment at a certain time falls within a certain process interval, the segment is determined to be in the corresponding process. If a segment spans multiple process intervals within a time window, the process with the longest dwell time is taken as the main process, and a cross-process marker is recorded. For example, in the pressing stage, the segment thickness is updated by shrinking based on the influence of pressure and temperature on thickness. , This is the updated thickness value (or updated status value) for the segment at the next time step. This represents the original thickness value (or current state value) of the segment at the current time step. λ is the attenuation or change coefficient of the state change during pressing or curing. Its value can be determined based on historical process data. For example, λ can be a real number between 0.05 and 0.2. The sampling interval or simulation step size is set between adjacent time steps. During the solidification phase, the temperature is updated by decay over time. This can be achieved by adopting a rule that gradually decreases the temperature over time, such as reducing the temperature by a fixed decay rate within each time step, or setting a temperature decrease rate range based on historical data (e.g., 1℃~3℃ per second). The system recursively calculates the state of each segment according to the process sequence, obtaining the state evolution results across processes and thus cross-process simulation data. Through segment continuity constraints, allowable ranges are set for thickness differences between adjacent segments. The spatial location, thickness, temperature, and density information of each segment are uniformly mapped into field data, and different physical quantities are normalized. A transition region is constructed at the segment boundary, and the states of adjacent segments are smoothly merged using a weighted method based on spatial distance. , This represents the thickness value at position s after fusion. This represents the thickness value of the segment to the left of the boundary at the corresponding location. The weight function corresponds to position s, and its value ranges from 0 to 1. The spatial coordinates within the fusion zone This represents the thickness value of the segment to the right of the boundary at the corresponding location. By sequentially splicing and merging all segments, the overall field distribution data of the continuous slab is reconstructed.

[0032] Step S4: Perform continuous twin representation based on the overall field data of the continuous slab to obtain slab twin data.

[0033] In one embodiment, the system constructs a twin representation based on the overall field data of a continuous slab. The overall field is spatially discretized into multiple grid nodes, and each node is bound with corresponding state information, including parameters such as thickness, temperature, and density. The density is generated as follows: during the pressing process, the system calculates density changes based on the thickness variation of the section and the mass conservation relationship; that is, under the condition that the raw material mass remains approximately constant, a decrease in thickness corresponds to an increase in density. During the curing process, the density is stabilized and corrected based on temperature changes and material curing characteristics, gradually bringing it towards a stable value. In the absence of direct sensor measurement, the density can be estimated jointly by thickness and process state, and updated synchronously with the time history footprint of the section. A topological connection structure is constructed based on the spatial proximity relationship between nodes. The state of each node is recursively updated according to a preset time step, and the update rule is provided by the aforementioned simulation model. During the update process, abnormal nodes (such as those with state values ​​exceeding a set range) undergo neighborhood consistency correction. For example, when the state value (such as thickness, temperature, or density) of a grid node exceeds a preset reasonable range (e.g., thickness deviates from the neighborhood average by more than ±10%, or temperature deviates by more than ±15℃), the system marks the node as an abnormal node. It selects several adjacent nodes (e.g., two nodes before and after the node) to form a neighborhood set and calculates the average or median value of the corresponding state within the neighborhood as a reference benchmark. Based on this benchmark, the state value of the abnormal node is corrected to return it to the allowable range within the neighborhood, while maintaining its original trend direction. For multiple consecutive abnormal nodes, a step-by-step smoothing process is used for segmented correction to avoid new abrupt changes. Discrete node data undergoes continuous reconstruction processing, generating a continuous spatial state distribution through interpolation or smoothing fitting. The system obtains slab twin data containing spatial location, state distribution, and temporal evolution information.

[0034] Optionally, step S1 specifically includes: Step S11: Obtain continuous slab flow data; In one embodiment, the system synchronously collects continuous slab flow data during the feeding, laying, pressing, curing, and conveying stages of the insulation board continuous production line. The data includes at least the conveyor belt speed, slab thickness, slab surface temperature, and a sequence of apparent images acquired by a line scan camera, along with corresponding timestamp information. The system receives and aggregates various types of sensor data according to a preset acquisition cycle, retaining the original sampling frequency and equipment identification information of each data source to obtain continuous slab flow data.

[0035] Step S12: Perform heterogeneous sensing time-domain normalization on the continuous slab flow data to obtain heterogeneous sensing time-domain data; In one embodiment, to address the issue of inconsistent sampling frequencies among different sensing devices, the system sets a unified time step as the normalization granularity and constructs a standard time axis accordingly. For thickness and temperature data, interpolation between adjacent sampling points is used to map the original data to the standard time axis; for image sequence data, the image frame closest to the target time is selected as the corresponding result. When there is no valid data before or after a target time, that time is marked as a missing measurement point. The original timestamps are compensated and corrected by combining the fixed communication delay parameters of each sensing device, enabling synchronous alignment of multi-source data under a unified time axis. After the above processing, a multi-source sensing record containing conveying speed, thickness, temperature, and image information is formed at each target time, thus obtaining normalized heterogeneous sensing time-domain data.

[0036] Step S13: Generate virtual segments from the heterogeneous sensing time-domain data to obtain virtual segment data; In one embodiment, the system establishes a correspondence between time and spatial displacement based on conveyor belt speed data. By calculating the cumulative change of speed over time, the time-series data is mapped to a spatial position sequence. The continuous slab is divided into segments according to a preset displacement length threshold. When the cumulative displacement between adjacent time nodes reaches or exceeds this threshold, a new virtual segment is generated. Each virtual segment corresponds to a defined spatial range, and its covered time interval is recorded simultaneously. During segment generation, the system performs statistical processing on the data within the segment, extracting attribute information such as average thickness, average temperature, image frame index, and average speed. When there are missing measurements within a segment, the corresponding missing measurement marker is retained instead of removing the segment. Through the above processing, the continuous slab flow data is transformed into virtual segment data with clear spatial boundaries and attribute characteristics.

[0037] Step S14: Anchor the virtual and real spaces based on the virtual segment data to obtain the sensing segment data.

[0038] In one embodiment, the system matches the spatial range corresponding to a virtual segment with the actual spatial coordinates of each workstation in the production line to establish a correspondence between the virtual segment and its physical location. Spatial boundary information for workstations such as the laying area, pressing area, curing area, and conveying area is pre-acquired. When the center of a virtual segment falls within the range of the corresponding workstation, the segment is anchored to that workstation. By combining edge change features, abrupt thickness changes, or temperature change areas in the image, the boundaries of the virtual segments are locally corrected to better reflect the actual position of the slab on the production line. For example, after the initial boundary of the virtual segment is determined, the system performs joint analysis on multi-source data within the neighborhood of that boundary (e.g., ±0.2m before and after the boundary, or corresponding to 3-5 sampling points). For visual image data, edge detection is performed on the image within the corresponding interval to extract locations of abrupt changes in grayscale or texture, and the location with the most significant change in edge intensity is designated as the boundary point. For thickness data, the difference between adjacent sampling points is calculated, and when the thickness change exceeds a preset threshold (e.g., a change exceeding 5%), the location is marked as a thickness abrupt change point. For temperature data, regions with significant temperature gradient changes (e.g., a temperature difference between adjacent points exceeding 8℃) are identified as temperature change boundary references. The system aligns the above three types of candidate boundary points. When multiple features appear near the same spatial location (e.g., an error not exceeding 0.1m), that location is selected as the priority correction boundary; if only a single feature meets the condition, its change intensity is considered for selection. The original virtual segment boundary is offset and adjusted towards the selected candidate boundary position, making the corrected segment boundary more closely match the actual slab's state change boundaries at different workstations. After anchoring, each virtual segment includes a segment identifier, spatial range, workstation, and multi-source state attributes, thus forming perceptual segment data that corresponds to the actual production location.

[0039] Optionally, step S2 specifically includes: Step S21: Extract segment anchor points based on the sensing segment data to obtain segment anchor point data; In one embodiment, the system performs statistical analysis and feature extraction on the data of each sensing segment. The center moment of the segment is selected as the time anchor point, and the midpoint of the segment's spatial range is selected as the spatial anchor point. At this time position, the average thickness within the segment is calculated, and the temperature change trend is obtained by calculating the difference between adjacent sampling points. Simultaneously, the image feature vector at the corresponding moment is extracted as appearance description information. The system calculates the thickness difference between the beginning and end moments of the segment. The time anchor point, spatial anchor point, and the aforementioned multi-source state features are combined to form segment anchor point data.

[0040] Step S22: Perform multimodal inertial fingerprint encoding on the segment anchor point data to obtain segment inertial fingerprint data; In one embodiment, the system uses segment anchor point data as a basis and analyzes multi-source state changes within a sliding time window set before and after the segment's center time to construct the segment's inertial fingerprint features. The system calculates the thickness variation amplitude and temperature variation amplitude between adjacent time points; simultaneously, it obtains the apparent change degree / apparent features by comparing the differences in image features (such as texture features) between consecutive time points. The state data within this time window is statistically analyzed to obtain the thickness or temperature fluctuation range. The average thickness, temperature change trend, the aforementioned variation amplitude, fluctuation range, and image features are combined to form the segment's inertial fingerprint data.

[0041] Step S23: Based on the segment inertial fingerprint data, perform segment continuity probability deduction to obtain segment continuity relationship data; In one embodiment, the system extrapolates the continuity of segments in subsequent time windows based on segment inertial fingerprint data. Based on the current segment's spatial location and transport speed, it predicts its theoretical location range in the next time window and selects a set of segments within that range, such as... , This refers to the predicted spatial location (or predicted location coordinates) of the segment in the next time window. This represents the coordinates of the center spatial location of the current segment (or the center location of the segment). The speed of the conveyor belt at time t (or the instantaneous conveying speed). This refers to the time interval (or time step) between adjacent time windows. For each segment, the degree of difference between it and the current segment in terms of thickness, temperature, and surface texture is calculated, and its continuity is determined based on preset thresholds. For state changes caused by processes such as pressing or curing, the system sets corresponding compensations, reasonably relaxing the tolerances for thickness or temperature changes. When multiple segments exist, they are filtered based on temporal and spatial order constraints. The continuity relationship between segments is determined, and its degree of continuity is quantified to form segment continuity relationship data.

[0042] Step S24: Based on the segment continuity relationship data, perform cross-process fracture segment suturing to obtain segment suturing data; In one embodiment, the system identifies the location of interruption in the continuity of segments and extracts the segment pairs before and after the break. Based on the time interval between the two segments and the types of processes they cross, a reasonable displacement range and state change range are set. Forward state extrapolation is performed on the segment before the break, and reverse backtracking analysis is performed on the segment after the break. Within the intermediate interval, the continuity of the state change trends of the two segments is determined, such as whether the direction of change is consistent or whether the magnitude of change is within the allowable range. For example, the system uses the state of the segment before the break at the moment of interruption as the initial condition, and performs forward recursive calculations on thickness, temperature, and density according to the state transition rules corresponding to its current process to obtain a predicted state sequence within the interruption interval. Starting from the state of the segment after the break at the beginning, backtracking calculations are performed in the reverse direction of time. A backtracking state sequence is generated according to the reverse change rules of the corresponding process (e.g., temperature backtracking shows an increasing trend, thickness backtracking shows a recovering trend). The system selects several aligned position points (e.g., 2-5 equally spaced sampling points) within the interruption interval and compares the differences in thickness, temperature, and other states at each position between the forward extrapolation results and the reverse backtracking results. When the difference between the two segments is less than a preset threshold (e.g., thickness difference less than 5%, temperature difference less than 10℃) at most locations, and the direction of change is consistent (e.g., both showing a decreasing or increasing trend), it is determined that the two segments have a continuous evolution relationship / meet the continuity condition within this interval. When the continuity condition is met, a bridging segment is generated to connect the two segments. Its state can be constructed through linear change or by combining the change rules of the corresponding process. After bridging is completed, the system performs a consistency check on the adjacent segments of the stitching result to ensure that the original temporal order and spatial relationship are not disrupted. For example, the system checks whether the order of the stitched segment sequence in the time dimension remains monotonically increasing, that is, the end time of the previous segment should be less than or equal to the start time of the next segment; it checks whether the spatial position remains continuous, that is, the spatial position difference between adjacent segments should not exceed a preset displacement threshold (e.g., 0.1m~0.2m). The system verifies the state difference between the stitched segment and its adjacent segments before and after it to ensure that there are no abnormal abrupt changes in the state such as thickness and temperature at the connection point (e.g., thickness jump does not exceed ±8%). If a suture relationship causes adjacent segments to cross in sequence or abruptly change state beyond the acceptable range, the suture is deemed invalid and either discarded or rematched. The system establishes suture connection relationships between segments, obtaining segment suture data.

[0043] Step S25: Perform multi-scale state history expansion on the segment stitching data to obtain time-history footprint data.

[0044] In one embodiment, the system performs time-unfolding processing on the stitched segment links to obtain a segment state sequence. At the microscale, the system records the state changes of each segment, such as thickness, temperature, and appearance features (texture features extracted from images), step by step over time. At the process scale, the system summarizes the state of the segments in different process stages to obtain the average or typical state features corresponding to each process. For example, the system divides the segment state sequence into multiple process sub-sequences based on the process label to which the segment belongs, such as a paving stage sequence, a pressing stage sequence, and a curing stage sequence. For each process sub-sequence, the system performs statistical summarization processing on the state data of all segments within that stage. For continuous state quantities (such as thickness, temperature, and density), the average or median value within that process stage can be calculated as a representative state; for example, the thickness in the pressing stage can be the average thickness of all segments within that stage. For state quantities with large fluctuations, the interval value with high frequency or the average value of a stable stage can be selected as a typical state. The system extracts the typical trend of state changes within that process stage; for example, the temperature gradually decreases during the curing stage. At the overall scale, the segments are connected in temporal and spatial order to construct an evolution curve covering the entire length of the slab. The segment links are organized into a directed sequence according to their order, and each segment is assigned corresponding temporal, spatial location, and multi-source state attributes. The system obtains time-history footprint data that simultaneously contains temporal evolution, spatial distribution, and multi-dimensional state information.

[0045] Optionally, the segment continuation probability deduction is as follows: Segment inertial kernels are extracted from the segment inertial fingerprint data to obtain segment inertial kernel data; In one embodiment, the system performs stability analysis and screening of each feature component based on segmental inertial fingerprint data. The thickness change trend, temperature change trend, apparent texture features, and their fluctuations are analyzed item by item. Components whose change direction remains consistent and whose fluctuation amplitude is within a preset range over multiple consecutive time windows are selected as stable features. For example, the change direction is calculated by differencing adjacent time points. For instance, when the thickness shows a decreasing trend over consecutive time windows (i.e., the difference results are all negative), the change direction is considered consistent. Temperature and texture features are similarly judged for directional consistency using difference or feature distance changes. The fluctuation amplitude of each component within the window is calculated, for example, by the difference between the maximum and minimum values ​​or the standard deviation. When the thickness fluctuation does not exceed ±5% of its average value and the temperature fluctuation does not exceed ±8℃, the component is considered to be within a stable range. Only when a feature simultaneously meets both the conditions of consistent change direction and fluctuation amplitude less than a threshold is it selected as a stable feature for constructing the segmental inertial kernel. For components with large fluctuations, only trend information such as the direction of change and the location of extreme values ​​is extracted, without retaining instantaneous values. For example, the overall direction of change of the component in the time series (rising, falling, or rising then falling, etc.) is calculated, and the time position of its maximum or minimum value (such as the second or fourth time window) is recorded as a representative feature of the component. The trend of change is encoded into simple identifiers (e.g., "+1" indicates rising, "-1" indicates falling, and "0" indicates fluctuating). The selected thickness stability features, temperature response stability features, and texture continuity features are combined to form segmental inertial kernel data.

[0046] The forward reachability domain is expanded based on the segment inertial kernel data to obtain the forward reachability domain data; In one embodiment, the system predicts the possible reach of the current segment in the next time window based on the spatial center position and the corresponding transport speed, thus obtaining the theoretical position. Based on factors such as speed fluctuations, local stretching or compression, and boundary offsets during transport, a certain spatial tolerance range is pre-set around the theoretical position, forming a position interval, such as... , The predicted spatial location of the segment within the next time window. This refers to the spatial tolerance range set around the predicted location. In addition to spatial constraints, the system also determines the corresponding allowable range of state changes based on the thickness and temperature variation trends within the segment's inertial core. For example, the thickness is allowed to shrink moderately during the pressing stage, and the temperature is allowed to gradually decrease according to the process rules during the curing stage. The aforementioned spatial location range and state change range are jointly expressed to form the forward reachable domain data.

[0047] Multi-channel preservation of forward reachable domain data is calculated to obtain continuity preservation data. In one embodiment, the system extracts segments within the forward reachable domain and performs a retention evaluation on each segment from multiple feature channels. The thickness difference between candidate segments and the current segment is calculated to determine thickness retention; temperature retention is evaluated based on the consistency of temperature change directions; the retention of apparent texture is evaluated by comparing the degree of difference between image features or texture encodings; and the continuity of boundary contours is determined by considering the change direction of the segment's beginning and end boundaries. Segments that meet a preset tolerance range in most feature channels are deemed to have a high probability of continuity. The system generates corresponding multi-channel retention evaluation results for each candidate segment, forming continuity retention data.

[0048] Based on the continuity retention data, inertial decay compensation is extrapolated to obtain compensation retention data; In one embodiment, the system performs process-based correction processing on the continuity retention results based on the process type that the segment is currently in or about to enter. Under the pressing process, shrinkage compensation is set for thickness retention, allowing a reasonable decrease in section thickness relative to the previous moment, but restricting abnormal increases. Since the physical essence of the pressing process is the compaction of material under pressure, the direct result is a decrease in thickness and an increase in density. Therefore, in the continuation judgment, a certain degree of decrease in candidate section thickness relative to the previous moment is allowed, for example, a reduction in thickness within the range of 5% to 15%. Abnormal increases in thickness do not conform to the pressing process rules and originate from mismatches or data anomalies. Therefore, a strict limit is set on thickness increases (e.g., not exceeding ±2%). Exceeding this range reduces its continuation reliability or directly determines it as discontinuous. Under the curing process, thermal decay compensation is introduced for temperature retention, allowing the temperature to gradually decrease according to the process rules. In the curing process, the slab undergoes a gradual cooling process from high temperature to ambient temperature. Therefore, the temperature change has a monotonically decreasing or slowly decreasing characteristic. When making continuation judgments, the system allows a reasonable decrease in candidate section temperature relative to the previous moment, for example, a temperature decrease of 1℃ to 5℃ within each time window. However, if a significant reverse temperature increase occurs (e.g., exceeding +5℃), it is considered an abnormal change. Meanwhile, for surface texture features, a certain degree of weakening is permissible in high-temperature or compaction environments, but the overall distribution pattern must remain continuous and without abrupt changes. That is, during high-temperature curing or pressing, the surface texture features of the slab may weaken due to material softening or compaction, for example, reduced edge contrast or blurred texture details. Therefore, in the continuity judgment, a decrease in texture feature intensity is allowed within a certain range, for example, a decrease of 10% to 20%; however, the overall distribution structure of the texture (such as the main outline or regional distribution) must remain consistent, without significant morphological changes or abrupt changes. Through the above compensation, normal and abnormal changes caused by the process are distinguished, resulting in corrected retention results and forming compensated retention data.

[0049] Based on the compensation preservation data, neighboring segment cooperative exclusion inference is performed to obtain neighboring segment competition relationship data; In one embodiment, the system performs neighbor-segment cooperative exclusion inference when multiple preceding segments simultaneously match the same segment. The system detects whether there is overlap between the current segment and its preceding and following adjacent segments in the forward reachable domain; when there is overlap, it performs consistency evaluation on each pairing relationship, including whether the temporal and spatial order of the segments is maintained, whether the segment boundary changes are continuous (the continuity of boundary changes can be judged by the difference between adjacent segment boundaries, such as the change in thickness or temperature at the boundary does not exceed a preset range (e.g., thickness change ≤ ±8%, temperature change ≤ ±10℃)), and whether the state evolution is smooth (whether the state evolution is smooth is judged by the changing trend of the state difference of continuous segments, requiring that the change in adjacent difference does not show a sudden increase or decrease (e.g., the difference change does not exceed twice that of the previous moment)). If a pairing relationship leads to an intersection of adjacent segments, a reversal of boundaries, or a sudden change in state (a pairing relationship is considered an intersection if the time or spatial position of a subsequent segment is earlier than its preceding segment; a boundary reversal if the segment boundary position changes in the opposite direction (e.g., the starting position of the subsequent segment is more than a certain threshold shorter than the ending position of the preceding segment by 0.05m); a state change if the change in the state components of adjacent segments exceeds a preset jump threshold (e.g., a change in thickness exceeding ±12% or a change in temperature exceeding ±15℃), then the priority of the pairing relationship is reduced or it is eliminated. Through the above exclusion and filtering processes, the system retains a set of non-conflicting relationships, forming adjacent segment competition relationship data.

[0050] Based on the data on the competition relationship between adjacent segments, discontinuous crossing and bridging simulation is performed to obtain crossing and bridging data; In one embodiment, when a segment fails to find a matching continuation segment within the next time window, the system does not directly determine that the link has terminated, but continues the continuation search in subsequent time windows. Based on the inertial kernel characteristics and cumulative displacement estimation results of the segment, an reachable range across time windows is constructed, and candidate segments are retrieved within this range. For sections that, although spanning one or more time windows, still meet the requirements for state continuity, the system assesses whether changes in thickness, temperature, and texture conform to the process evolution pattern. It also determines whether the cumulative displacement deviation is within a preset range. For example, in the pressing process, the thickness is required to decrease within a reasonable range (e.g., a decrease of 5% to 15%), while the density is required to increase or stabilize. In the curing process, the temperature is required to gradually decrease (e.g., a decrease of 1°C to 5°C per time window), with the direction of change continuous and not reversed. For texture features, the system judges based on feature similarity or structural consistency, requiring a consistent overall distribution pattern, allowing only a weakening of intensity within a certain range (e.g., a decrease of 10% to 20%), and prohibiting structural abrupt changes. Regarding displacement, the system calculates the theoretical displacement value based on the initial position and conveying speed of the section, obtains the predicted position across windows, and compares it with the actual position of the section. When the deviation does not exceed a preset range (e.g., ±0.1m to 0.3m), it is determined that the displacement consistency condition is met. Only when the state evolution conforms to the process pattern and the displacement deviation is within the allowable range will the system establish a discontinuous crossing bridging relationship. When the above conditions are met, a bridging relationship for discontinuous crossings is established. The system then obtains the crossing bridging data.

[0051] Based on the bridging data, the continuity confidence is aggregated to obtain the segment continuity probability data; In one embodiment, the system summarizes multi-dimensional information on the continuity relationship of each candidate segment, combining multi-channel retention results, process compensation results, adjacent segment cooperative exclusion results, and discontinuous bridging results to form a corresponding continuity credibility. The system determines whether the relationship simultaneously satisfies basic conditions such as spatial continuity, continuous state change, and consistent order of adjacent segments, and only retains relationships that meet the above conditions. For the retained relationships, they are classified into different credibility levels based on factors such as the number of feature channels they satisfy, whether there is cross-time window bridging, and the magnitude of state deviation after compensation. For example, when classifying the credibility level of the retained continuity relationships, the system counts the number of feature channels they satisfy, such as thickness retention, temperature retention, texture retention, and boundary continuity, a total of four channels. If a relationship simultaneously satisfies four of the following conditions, with no cross-time window bridging, and the compensated thickness deviation does not exceed 3% and the temperature deviation does not exceed 5℃, it is classified as a high-confidence level. If it satisfies three of the following conditions, with one cross-time window bridging, and the compensated thickness deviation does not exceed 5% and the temperature deviation does not exceed 8℃, it is classified as a medium-confidence level. If it only satisfies two of the following conditions, or although it satisfies three of the following conditions, there are more than two time window bridgings, and the compensated thickness deviation reaches 8% and the temperature deviation is within 12℃, it is classified as a low-confidence level. If the conditions are not met, such as only one channel being satisfied or the compensated deviation exceeding the allowable range, it will be directly rejected and not included in subsequent continuation relationships. Each confidence level is mapped to a corresponding continuation probability interval (high confidence level is mapped to a continuation probability interval of 0.85~1.00, medium confidence level to 0.60~0.85, and low confidence level to 0.40~0.60). The system selects the feature with the higher value among its retention or continuity indicators as the dominant feature (e.g., the larger value in thickness retention or temperature retention) and normalizes this feature to a score value in the range of 0~1. The system then performs a linear mapping on this score value according to the upper and lower bounds of the corresponding confidence level interval to obtain the specific continuation probability, forming segment continuation probability data.

[0052] The main continuation chain is extracted based on the segment continuation probability data to obtain segment continuation relationship data.

[0053] In one embodiment, the system extracts the main continuation chain based on segment continuation probability data. In the continuation relationships of each segment, the candidate segment with the highest continuation probability and satisfying overall temporal and spatial order consistency is preferentially selected as the main continuation target. When there is a conflict between candidate targets between adjacent segments, the system compares the continuation probability and boundary continuity of each candidate relationship, prioritizing the retention of relationships with higher probabilities and better continuity, and marking other relationships as alternatives. The system sequentially executes the above selection process along the time axis, gradually connecting the optimal continuation relationships of each segment to form a globally consistent main continuation chain. The system outputs segment continuation relationship data, including the identifiers of the preceding and following segments, the corresponding continuation probabilities, whether the connection was established through bridging, and information such as adjacent segment consistency markings.

[0054] Optionally, the cross-process fracture suturing specifically involves: Breakpoint pairing is performed based on segment continuity data to obtain breakpoint pairing data; In one embodiment, the system performs preliminary screening of candidate breakpoints within a preset time window based on time intervals, theoretical displacement estimated by conveying speed, and process span, obtaining breakpoint pairing relationships that satisfy basic spatiotemporal continuity. The system performs traversal analysis of the link based on segment continuity data, identifying segments without subsequent connections as forward breakpoints and segments without preceding connections as backward breakpoints. For each forward breakpoint, the system searches for backward breakpoint candidates within its subsequent preset time range, and filters them based on time intervals, theoretical displacement changes accumulated from conveying speed, and the range of processes crossed. When the actual spatial location of the backward breakpoint matches the theoretical displacement prediction range, and the number of processes it crosses does not exceed a preset limit, a corresponding breakpoint pairing relationship is established. The system obtains breakpoint pairing data, including forward and backward breakpoint identifiers, corresponding time spans, displacement spans, and information on the processes involved.

[0055] The process traversal interpretable domain is constructed from the breakpoint pairing data to obtain joint interpretable domain data; In one embodiment, for each breakpoint pairing relationship, the system constructs a joint interpretation domain including time constraints, spatial reachability constraints, and process state evolution constraints. This domain represents the realizability boundary of the pairing relationship under cross-process conditions. The time constraint is defined by the equipment operating cycle time, the spatial constraint consists of theoretical displacement and dynamic tolerance, and the state constraint is determined by process evolution rules. Based on the time interval between preceding and following breakpoints and the equipment operating cycle time, an allowable time range is determined. The time interval between preceding and following breakpoints refers to the time difference between the end time of the forward breakpoint and the start time of the backward breakpoint. The equipment operating cycle time refers to the average time required for the corresponding process equipment in the production line to complete one standard processing action or advance the slab to a fixed station, which can be obtained from historical operating data recorded by the equipment control system. Based on the theoretical displacement change and existing local offsets during the conveying process, the spatial displacement range is determined. The system calculates the theoretical displacement change based on the conveying speed sequence within the time interval between breakpoints, for example, through... ( This represents the theoretical displacement change between the breakpoints. The start time of the backward breakpoint. The time when the forward breakpoint ends. Let be the instantaneous speed of the conveyor belt at time t. The theoretical displacement that the forward breakpoint should reach under undisturbed conditions is obtained by using time as the variable. Due to factors such as conveyor belt speed fluctuations, local tensile and compressive stresses on the slab, and position detection errors, a local offset tolerance is pre-set based on this theoretical displacement to construct a spatial displacement range. For example, when the calculated theoretical displacement is 1.20m, the allowable offset can be set to ±0.15m according to the working conditions, thus obtaining a spatial displacement range of [1.05m, 1.35m]. If there is a significant compression or deviation phenomenon in the corresponding process, it can be appropriately expanded to ±0.20m. The system compares the actual displacement change of the backward breakpoint with this range. If it falls within this range, it is considered to meet the spatial interpretability requirements. Combining the process characteristics of the processes spanned, the allowable range of state changes is set. For example, in the pressing process, a reasonable reduction in thickness is allowed while an abnormal increase is restricted; in the curing process, a gradual decrease in temperature and a tendency for the state to stabilize are allowed. The system unifies the three types of constraints—time, displacement, and state—to form a joint interpretation domain.

[0056] Based on the joint interpretation domain data, a two-way evolutionary docking simulation is performed to obtain docking simulation data; In one embodiment, the system performs bidirectional evolutionary deduction of the breakpoint pairing relationship based on a joint interpretation domain. Starting from the end state of the segment corresponding to the previous breakpoint, the system performs forward deduction according to the state change rules of the processes crossed, generating a forward state sequence within the interrupted interval. For example, the system uses the state vector of the segment corresponding to the previous breakpoint at the start of the interruption as initial input. This state vector includes parameters such as thickness, temperature, and boundary position. The system recursively calls the corresponding state change rules according to the types of processes crossed between the breakpoints. For instance, when crossing a pressing process, the thickness state is updated according to the rule of gradually decreasing thickness and gradually increasing density, with the thickness shrinkage rate set to 2%~5% of the previous thickness in each time step. When crossing a curing process, the temperature state is updated according to the rule of gradually decreasing temperature with a rate of change that is initially fast and then slows down, with a temperature decrease of 1℃~4℃ in each time window. For the boundary position, the cumulative displacement based on the conveying speed is used for synchronous advancement. By updating each state sequentially according to time steps, a forward state sequence within the interrupted interval is obtained. Simultaneously, starting from the initial state of the segment corresponding to the breakpoint, backtracking is performed according to the reverse process rules to generate a backward state sequence within the interrupted interval. For example, the state of the segment corresponding to the breakpoint at the end of the interruption is used as the backtracking starting point, and state recovery calculations are performed in the reverse direction of time. The reverse process rules correspond to the forward rules. For example, if the pressing process is crossed, the thickness is allowed to change with a small recovery trend during backtracking, that is, the thickness increases by 2% to 5% at each time step to simulate the state before pressing; if the curing process is crossed, the temperature gradually rises according to the reverse temperature change rules during backtracking, for example, it rises by 1℃ to 4℃ at each time window until it approaches the exit state of the previous process. For boundary positions, the backward cumulative displacement is pushed back according to the conveying speed. Through the above reverse update, a backward state sequence within the interrupted interval is generated. The system performs consistency comparisons on the two sequences at various intermediate moments, assessing whether the directions of thickness changes, temperature changes, and boundary changes remain coordinated. For example, within the interruption interval, the system selects several alignment moments or equally spaced points and compares the forward and backward state sequences item by item. For thickness changes, it determines whether the directions of change of the two sequences at corresponding positions are consistent, such as whether they both show a decreasing or recovering trend. For temperature changes, it determines whether the two sequences both maintain a consistent decreasing or rising trend, and whether the corresponding state differences do not exceed preset thresholds (e.g., thickness difference not exceeding 5%, temperature difference not exceeding 8℃). For the direction of boundary changes, it compares whether the advancing directions at the boundary positions are consistent, such as whether they both extend continuously along the conveying direction without any reverse jump. When the two sequences have continuous overlapping intervals within the range defined by the joint interpretation domain, it is determined that the pairing at the breakpoint has the conditions for effective docking. The system records the start and end positions, corresponding times, and degree of state continuity of the overlapping intervals, forming docking simulation data.

[0057] Based on the docking simulation data, implicit bridging segments in the interruption zone are generated to obtain bridging segment data; In one embodiment, the system generates implicit bridging segments within the corresponding interruption intervals for breakpoint pairings that meet the docking conditions. Using the overlapping intervals of the forward deduction sequence and the backward backtracking sequence as constraint boundaries, a transition sequence of intermediate states is constructed, ensuring the bridging segments are spatially continuous and conform to the evolutionary patterns of the processes they span in terms of thickness, temperature, and density changes. For example, when the interruption interval for a certain breakpoint pairing is 2 seconds, corresponding to a spatial span of 1.2m to 1.5m, the system generates 4 to 5 bridging nodes within this interval, divided by 0.3m intervals, and constructs state values ​​at each node. In the pressing process, the thickness can be generated by decreasing by approximately 2% to 4% per node, and in the curing process, the temperature can be transitioned by decreasing by 2°C to 3°C per node. Simultaneously, the two ends of the bridging segment are aligned with the end point of the forward deduction sequence and the start point of the backward backtracking sequence, respectively. The generated bridging segments include at least the corresponding spatial start and end positions, time range, and multi-source state description information. When the interruption involves multiple processes, the system generates bridging states in segments according to the process sequence, maintaining the continuity of state transmission at the boundaries between processes. The generated bridging segments do not replace the original acquired data, but serve as a compensation structure for the interrupted area.

[0058] Based on the bridging segment data, neighboring segment collaborative closure verification is performed to obtain segment stitching data.

[0059] In one embodiment, the system inserts the generated bridging segment into the original segment link and performs adjacent segment collaborative closure verification on the stitching result. The system sequentially checks the consistency of the stitched segment with its preceding and following adjacent segments in terms of temporal order, spatial adjacency, and state change trend. This requires that the segment links maintain a non-overlapping sequence (the end time of any preceding segment is no greater than the start time of its subsequent segment), no spatial overlap (the spatial intervals of adjacent segments do not overlap in reverse, and their start and end positions meet a preset error range), and that adjacent segments exhibit continuous and smooth changes in thickness, temperature, and other state components, without abnormal reverse jumps (the difference in thickness, temperature, and other state components between adjacent segments does not exceed a preset threshold, and the direction of change remains consistent; for example, the thickness difference does not exceed ±5%, the temperature difference does not exceed ±8℃, and adjacent difference changes do not exhibit abrupt reverse changes (i.e., the change sign does not change abruptly)). If the introduction of the bridging segment causes link interleaving or local state discontinuity, the system cancels the stitching relationship; when the closure conditions are met, the bridging segment is confirmed as valid, and a continuous connection is established between the preceding breakpoint segment, the bridging segment, and the following breakpoint segment. The system obtains segment suture data, including suture link identifiers, bridging segment identifiers, cross-process information, and closure verification results.

[0060] Optionally, cross-process heterogeneous coupling simulation specifically includes: Step S31: Locate the process segment in the time-travel footprint data to obtain the process segment data; In one embodiment, the system matches the timestamps, spatial ranges, and migration trajectory information of each segment in the time-travel footprint data with a pre-stored production line process boundary table to determine the process stage of each segment at each moment. The production line process boundary table includes the spatial start and end ranges of areas such as the laying area, pressing area, curing area, and pre-cutting conveying area. When the center position of a segment falls within the spatial range of the corresponding process at a certain moment, the segment at that moment is marked as that process type. If a segment crosses multiple process boundaries within the same time window, it is judged based on its dwell time in each process, with the process with the longer dwell time being designated as the main process, and the boundary crossing is marked. The system establishes a correspondence between segment identifiers, time ranges, and process labels to form segment process data.

[0061] Step S32: Construct the segment state transition and continuous coupling constraints for the segment process data to obtain the state transition model and continuous coupling constraint data; In one embodiment, the system constructs state transition relationships for each process based on segment process data. In the paving process, the thickness, temperature, and conveying status of the segment are used as inputs to establish state update rules after paving. In the pressing process, based on the process characteristics of reduced thickness and increased density caused by pressure, an update relationship between thickness shrinkage and temperature changes is constructed. In the curing process, an evolution rule is established where temperature gradually decreases over time and density tends to stabilize. The system uses parameters such as thickness, temperature, and density as segment state descriptions and performs recursive calculations based on the corresponding state update rules for each process. , For the i-th segment in the next time step Update the state vector. This is the state transition function (or process state update model) for the corresponding process p. Let be the state vector (including state parameters such as thickness, temperature, and density) of the i-th segment at time t. Simultaneously, continuous coupling constraints are constructed for adjacent segments, requiring that differences in thickness, temperature, and density be controlled within preset ranges, such as... , Let be the thickness value of the i-th segment. Let be the thickness value of the (i+1)th segment. This is the allowable threshold for the thickness difference between adjacent sections (e.g., 0.5 mm or ±5%). Let i be the temperature value of the i-th segment. The temperature value of the (i+1)th segment. This is the allowable threshold for temperature difference between adjacent sections (e.g., 5℃~10℃). The system obtains the state transition model and continuous coupling constraint data (whether the specific definitions of the state transition model and continuous coupling constraint data need to be given).

[0062] Step S33: Perform cross-process state recursion calculations based on the state transition model and continuous coupling constraint data to obtain cross-process simulation data.

[0063] In one embodiment, the system uses the initial state of each segment in the time-travel footprint as the starting point for recursion and updates the state step by step according to the time sequence and the corresponding process sequence. At each time step, the system calls the corresponding state transition rule according to the process to which the current segment belongs to obtain the temporary state of the segment; this state is then checked for continuity constraints with adjacent segments. When the differences in state such as thickness, temperature, or density exceed a preset range, the relevant components are locally corrected to bring them back to the allowable range. For segments where process switching occurs, the output state of the previous process is used as the input of the next process to continue the recursive calculation until all processes are completed. The system obtains cross-process simulation data, including the state information of each segment at different times and under different processes, such as thickness, temperature, and density, as well as the corresponding process identifier.

[0064] Optionally, the reconstruction of the entire heterogeneous data field specifically includes: Based on cross-process simulation data, segment state field elements are extracted to obtain segment field data. In one embodiment, the system analyzes the segment state results in cross-process simulation data and extracts field element units representing local states. For each segment, its spatial range, process label, and corresponding thickness, temperature, and density are extracted. The center position of the segment's spatial range is used as the reference coordinate for the field element. The density state can be calculated from the pressing and curing processes in the aforementioned cross-process simulation stage. In the pressing process, the system calculates the change in segment density based on the initial thickness and the change in thickness after pressing, combined with the mass conservation relationship. For example, with the raw material mass remaining constant, a decrease in thickness corresponds to an increase in the material content per unit volume, thus obtaining the density change trend. In the curing process, the system performs stability correction on the density based on the temperature decay process and material curing characteristics, gradually bringing it towards a stable value. The spatial location, process information, and multi-dimensional state parameters are combined to form segment state field element data. When multiple sets of simulation states exist for a segment within the same time window, the state corresponding to the main process is used as the main description of the field element, while other states are additionally marked and retained. This process transforms the simulation results, originally organized in the form of a segment chain, into discrete field data with spatial positioning capabilities.

[0065] Based on the segment field metadata, heterogeneous state variables are mapped to the same domain to obtain the same domain field metadata; In one embodiment, the system constructs a homo-domain mapping for continuous slab field reconstruction, addressing the differences in dimensions, scale of change, and process semantics among state variables such as thickness, temperature, density, and structural characteristics in the segment field metadata. Using the set of neighboring segments for each segment as a local reference, the continuous state variables are relativized, converting the original state values ​​into offsets relative to the average level of the neighborhood. For example, the thickness state is represented as... ,in It represents the relative thickness offset (or thickness normalized expression value) of the i-th segment. This represents the original thickness value of the i-th segment. Let i be the set of neighboring segments of the i-th segment. For the index variable in the neighborhood segment set, Let be the thickness value of the neighboring segment j. Temperature and density states are processed in the same way, unifying different dimensional states into relative changes. The system combines the process labels to which the segment belongs to align state quantities under different process stages. For example, in the pressing process, a correspondence is established between thickness offset and density change, ensuring both are expressed in the same direction of change space; in the curing process, the temperature change trend is correlated with density stability, ensuring the mapped temperature and density components have consistent change characteristics under a unified semantic framework. Through these processes, state quantities with different physical meanings have a basis for collaborative computation within the same expression space. The system performs local consistency checks on the mapping results of adjacent field elements between segments. If the mapped state component of a certain segment deviates abnormally from its neighboring segments, it corrects the state based on the trend of the neighboring state changes to ensure that it meets the spatial gradient characteristics of the continuous slab. This correction process does not change the direction of the original state change, but only constrains the magnitude of the offset. For example, the system extracts the direction of the state change of each segment in the neighboring set of the segment, such as determining whether the overall thickness is increasing, decreasing, or stable. Then, it calculates the average or median magnitude of the change of the neighboring segments as a reference benchmark. When the state offset of the target segment is consistent with the direction of the neighboring trend but the magnitude is too large (e.g., deviating from the neighborhood average by more than ±10%), its offset magnitude is compressed to bring it back to the range of the neighborhood change (e.g., adjusted to within ±5%). When its direction of change is inconsistent with the overall trend of the neighborhood, its direction is corrected first to make it consistent with the dominant trend of the neighborhood, and then the magnitude is restricted. For example, when the thickness offset of a certain segment is +15%, while the average offset of its neighboring segments is +4% to +6%, the system can correct the thickness of that segment to around +5%. If the thickness offset of that segment is -10% while the average offset of its neighboring segments is +5%, the system first adjusts its direction to positive and then limits its amplitude to a reasonable range. For discrete information derived from visual or structural features, the system converts it into continuous intensity components using feature distance metrics, allowing it to participate in calculations within the same field representation framework as continuous physical quantities. The processed state components are then organized into co-domain field metadata according to a unified structure. Based on the same domain field metadata, the segment boundaries are gradually merged to obtain boundary fusion data; In one embodiment, to address the state discontinuity issue arising from adjacent segment field elements at boundary locations, the system establishes a fusion band of a certain range at the boundaries of each segment. This fusion band extends a predetermined distance to both sides of the segment boundary, such as... , The coordinates of the boundary center at the intersection of the sections. The half-width of the space extending to both sides of the boundary fusion zone (e.g., 0.05m~0.2m). Within the fusion zone, the system acquires the differences in thickness, temperature, and density between adjacent segments, and generates corresponding transition weights based on the distance relationship between the boundary position and the centers of the two adjacent segments, such as... , The value of the basic fusion weight function (or spatial transition weight) at position s. For exponential function operators, This is the distance attenuation coefficient (e.g., a positive number between 5 and 20). This represents the distance from position s to the center of the left segment. This represents the distance from position s to the center of the right segment; the system sets a state difference modulation term, for example, to dynamically adjust the weights based on thickness or temperature differences: , The value of the modulated fusion weight function at position s. The base fusion weight function value at position s. The modulation coefficient (e.g., 0.3~0.7). This refers to the state difference between the two sides of the boundary (such as thickness difference, temperature difference, or other difference indicators). A preset maximum difference value is used. This modulation makes the weight changes more gradual in boundary regions with large state differences. Gradual processing is applied to each state component within the fusion band, allowing the state value to smoothly transition from one segment to the other. When the state difference is large at the boundary, the system automatically expands the fusion band. After the above processing, boundary fusion data is obtained.

[0066] Based on the boundary fusion data, cross-process layered field splicing is performed to obtain the overall field data of the continuous slab.

[0067] In one embodiment, the system organizes field data after boundary fusion according to the spatial order of segments in the production line and performs layered splicing processing in conjunction with process labels. Field elements belonging to the same process stage are sequentially spliced ​​to form corresponding process sub-fields; each sub-field is connected sequentially according to the process flow of laying, pressing, and curing to construct the overall spatial field structure. During the splicing process, the state of the end segment of the previous process is used as a reference for the starting segment of the next process, and the continuity of states such as thickness, temperature, and density is checked. When there are local differences at the boundary between processes, local corrections are made based on the trend of changes in the neighboring state. For example, by calculating the state difference between adjacent segments within the neighboring region, the overall direction of change is determined (e.g., thickness gradually decreases, temperature gradually decreases, etc.), and the average change amplitude of the neighboring region is calculated as a reference benchmark. When the state of a boundary section deviates abnormally from the trend of its neighboring area (e.g., a sudden increase in thickness exceeding ±8% or a temperature jump exceeding ±10℃), the system first adjusts the direction of change to align with the dominant trend of the neighboring area, and then limits the magnitude of the change to the average range of the neighboring area (e.g., compressing it to ±3%~±5% or ±3℃~±5℃). For multiple consecutive sections with deviations, the system uses a step-by-step iterative correction to gradually transition the state back to the neighboring trend, rather than a one-time abrupt adjustment. Simultaneously, the inherent variation patterns of each process are preserved during the correction process; for example, the thickness decreasing trend is maintained during the pressing stage, and the temperature decreasing trend is maintained during the curing stage. The system obtains overall field data continuously distributed along the length of the slab, including the thickness, temperature, and density at each location.

[0068] Optionally, the gradual fading and merging of segment boundaries specifically refers to: Boundary fusion bands are constructed from the metadata of the same domain field to obtain boundary fusion band data; In one embodiment, the system identifies the boundaries of adjacent segments and determines their intersection positions based on the spatial order of each segment in the same domain data. The midpoint between the end position of the previous segment and the beginning position of the next segment is used as the boundary center, and a fusion zone is constructed on both sides of this position according to a preset length range. The width of the fusion zone can be adaptively adjusted based on the average length of the segments or the state fluctuations near the boundary. When there are significant thickness or temperature fluctuations, the fusion zone range is appropriately expanded to cover the transition area. For example, 20% to 30% of the average length of the segments can be used as the initial width (e.g., if the average length of the segments is 0.5m, the initial fusion zone width can be set to 0.1m to 0.15m). The system performs fluctuation analysis on the state data within a certain range near the boundary (e.g., 2 to 3 segments before and after), calculating the amplitude of thickness or temperature fluctuations, for example, by evaluating the difference between the maximum and minimum values ​​or the standard deviation. When the thickness fluctuation exceeds a preset threshold (e.g., ±5%) or the temperature fluctuation exceeds ±8℃, the boundary is determined to be a high-fluctuation region. In this case, the width of the fusion zone is increased proportionally (e.g., increased to 1.5 or 2 times the original width, i.e., 0.15m~0.3m). When the state change is relatively stable (e.g., thickness fluctuation less than ±3%, temperature fluctuation less than ±5℃), the width of the fusion zone is maintained or appropriately reduced (e.g., reduced to 0.08m~0.12m). The system organizes and records the corresponding left and right segment identifiers, spatial range, and related state information within the fusion zone to form boundary fusion zone data.

[0069] The difference between the two sides of the boundary is calculated based on the boundary fusion zone data to obtain the difference data between the two sides of the boundary; In one embodiment, the system performs a step-by-step difference analysis on the co-domain state components of the segments on both sides of the boundary based on boundary fusion zone data. Within the fusion zone, the system calculates the average differences in thickness, temperature, and density between the left and right segments, and assesses the consistency of the states on both sides accordingly. The system determines whether the state change trends on both sides of the boundary are consistent, for example, whether the thickness both show contraction or whether the temperature decreases synchronously. Boundaries with differences exceeding a preset range are marked as strong difference boundaries; otherwise, they are marked as weak difference boundaries. The system obtains difference data on both sides of the boundary, including the degree of difference of each state component and the corresponding boundary difference markers.

[0070] Distance difference transition data is obtained by calculating the distance difference transition based on the difference data on both sides of the boundary. In one embodiment, the system discretely sets multiple transition point locations along the spatial direction within the boundary fusion zone and calculates the positional relationship of each transition point relative to the boundary center. For each transition point, its spatial transition position attribute / transition position weight is determined by combining its relative distance to the center positions of the left and right segments. For example, the system assigns a transition position attribute / transition position weight to any transition point within the fusion zone. Calculate its position relative to the center of the left segment. and the center position of the right section The distance is obtained. and Normalized weights are constructed based on distance relationships, so that the closer the distance, the greater the weight. For example, the weight of the left side can be defined as... ,in and These represent the degree of influence of the left and right segments at the transition point, respectively, and satisfy the following conditions: When the transition point is close to the left segment Smaller Larger, therefore Approaching 1; conversely, when approaching the right segment, Approaching 1. The system can adjust the weights by combining transition intensity markers. For example, when the boundary is a strong difference boundary, a smoothing factor is introduced to buffer the weight changes, making the weights show a smoother trend in space (e.g., limiting the weight change of adjacent transition points to no more than ±0.1). At the same time, based on the boundary difference markers in the difference data on both sides of the boundary, the transition range and transition intensity are adaptively adjusted: when the boundary difference is small, a shorter transition interval is used to maintain local features; when the difference is large, the transition range is appropriately expanded. For example, when the thickness difference is less than 3% and the temperature difference is less than 5℃, it is judged as a low difference boundary, and the transition range is taken as a small value (e.g., 0.08m~0.12m), and the transition intensity is low; when the thickness difference reaches 5%~10% and the temperature difference reaches 8℃~12℃, it is judged as a medium difference boundary, corresponding to a moderate range (e.g., 0.12m~0.2m); when the difference increases, the transition range is expanded (e.g., 0.2m~0.3m) and the transition intensity is increased. Based on the above processing, descriptive information including location, left and right proximity, and transition intensity is generated for each transition point, forming distance difference transition data.

[0071] Boundary fusion data is obtained by performing bilateral coupling fusion calculations based on distance difference transition data.

[0072] In one embodiment, the system performs bilateral coupling fusion processing on each transition point within the fusion band based on distance difference transition data. At each transition point, the state information of the left and right segments is acquired, and the state gradually transitions from left to right according to the proximity of the transition point to the two segments and its fusion weight (i.e., the aforementioned transition position weight). For state components such as thickness, temperature, and density, adjustments are made continuously according to the transition position during the fusion process; when the corresponding boundary is determined to be a strong difference boundary, a gentler transition method is adopted to avoid abrupt changes at the boundary. After fusion, the system performs continuity verification on the overall state distribution within the fusion band and smooths out areas with local discontinuities. The system obtains boundary fusion data showing the continuous state distribution at each location within the fusion band.

[0073] Optionally, step S4 specifically includes: Step S41: Perform gridding mapping based on the overall field data of the continuous slab to obtain the slab mapping data; In one embodiment, the system performs spatial discretization on the overall field data of a continuous slab, dividing the thickness, temperature, and density states distributed along the length of the slab into multiple continuous grid cells according to a preset step size. Each grid cell corresponds to a defined spatial interval, with the center position of that interval serving as the grid node. The system maps the thickness, temperature, and density data within the corresponding interval in the overall field to that grid node, forming corresponding node state information. When multiple data points exist within a certain interval, a local averaging method is used to determine the node state; when data is missing within an interval, it is supplemented by data from adjacent intervals. This results in slab mapping data arranged in spatial order.

[0074] Step S42: Construct a twin structure from the slab mapping data to obtain twin structure data; In one embodiment, each grid node is defined as a basic unit node in the twin, corresponding to a spatial location unit of the slab in the actual production line. Each node is bound to its physical coordinates and corresponding thickness, temperature, and density state vectors, thus establishing a one-to-one mapping between virtual nodes and local areas of the actual slab. Structurally, the system constructs a multi-layered twin structure: in the geometry layer, the continuous spatial form of the slab is described through the connection relationships between adjacent grid nodes; in the state layer, the multi-dimensional state information of each node is bound to its corresponding node, representing the physical state distribution of the slab at different locations; in the process layer, nodes are divided into different process substructures according to their process labels, representing the structural distribution of the slab at each process stage; in the evolution layer, a temporal inheritance relationship is established for nodes at the same spatial location in different time slices, representing the evolution path of the slab state over time and as the process progresses. The system adds a state update interface to each node to receive the recursively calculated state changes. The connection relationships between nodes also carry state propagation constraints; when the state of a node changes, its influence can propagate along the connection relationships to adjacent nodes. The system organizes node sets, spatial connections, temporal evolution relationships, and process hierarchical relationships into multi-layered structured twin data.

[0075] Step S43: Recursively update the twin structure data to obtain the twin update data; In one embodiment, the system uses the current twin structure data as a basis and recursively updates the state of each grid node according to a preset time step. For each node, its current thickness, temperature, density, and other state information are read, and combined with the change trends obtained from cross-process simulation, the node state for the next time step is generated. After the update is completed, the system performs a continuity check on the state difference between the node and its neighboring nodes. When the local difference exceeds a preset range, the node is locally corrected according to the change trends of neighboring nodes to keep it consistent with the surrounding area. For nodes in abnormal areas, updates can be performed while retaining the abnormality marker. The system obtains twin update data containing the updated node states and their connection relationships.

[0076] Step S44: Perform continuous field reconstruction based on the twin update data to obtain slab twin data.

[0077] In one embodiment, the system reconstructs the continuous field distribution of the slab along its length based on the spatial location and corresponding state information of each node in the twin update data. Using the states of adjacent nodes as constraints, the system recovers the continuous variations in thickness, temperature, and density at non-node locations through interpolation or smooth fitting between nodes. For regions marked as anomalous, their local abrupt changes are preserved during reconstruction. After reconstruction, the system integrates the continuous field distribution results with node-level state information, topological connectivity, temporal evolution information, and process labels to form slab twin data.

[0078] Optionally, this application also provides a cloud data-driven twin construction system for insulation board production lines, used to execute the cloud data-driven twin construction method for insulation board production lines as described above. The cloud data-driven twin construction system for insulation board production lines includes: At least one processor; A memory, wherein the memory stores computer programs that can be executed by the processor; Communication interface, used for data interaction with production line sensing devices and cloud data platform; When the processor executes the computer program, it performs the following functions: The sensing section binding module is used to acquire continuous slab flow data; and to bind sensing sections to the continuous slab flow data to obtain sensing section data. The multidimensional time-history footprint reconstruction module is used to reconstruct the multidimensional time-history footprint based on the sensing segment data to obtain the time-history footprint data. The cross-process heterogeneous coupling simulation module is used to perform cross-process heterogeneous coupling simulation on the time-history footprint data to obtain cross-process simulation data; and to reconstruct the global heterogeneous data field based on the cross-process simulation data to obtain the overall field data of the continuous slab. The continuous twin expression module is used to continuously express twins based on the overall field data of the continuous slab, and obtain slab twin data.

[0079] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended application documents rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of the equivalents of the application documents be incorporated into the invention.

[0080] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A method for constructing a twin of an insulation board production line based on cloud data, characterized in that, Includes the following steps: Step S1: Obtain continuous slab flow data; bind sensing sections to the continuous slab flow data to obtain sensing section data; Step S2: Extract segment anchor points based on the sensing segment data to obtain segment anchor point data; perform multimodal inertial fingerprint encoding on the segment anchor point data to obtain segment inertial fingerprint data; perform segment continuation probability deduction based on the segment inertial fingerprint data to obtain segment continuation relationship data. Based on the segment continuity relationship data, cross-process fracture fragments are stitched together to obtain segment stitching data; multi-scale state history unfolding is performed on the segment stitching data to obtain time-history footprint data; Step S3: Perform cross-process heterogeneous coupling simulation on the time-history footprint data to obtain cross-process simulation data; reconstruct the global heterogeneous data field based on the cross-process simulation data to obtain the overall field data of the continuous slab. Step S4: Perform continuous twin representation based on the overall field data of the continuous slab to obtain slab twin data.

2. The method according to claim 1, characterized in that, Step S1 is as follows: Acquire continuous slab flow data; Heterogeneous sensing time-domain normalization is performed on continuous slab flow data to obtain heterogeneous sensing time-domain data. Virtual segments are generated from heterogeneous sensing time-domain data to obtain virtual segment data. Based on the virtual segment data, virtual and real space are anchored to obtain the perceived segment data.

3. The method according to claim 1, characterized in that, The specific deduction of the segment continuation probability is as follows: Segment inertial kernels are extracted from the segment inertial fingerprint data to obtain segment inertial kernel data; The forward reachability domain is expanded based on the segment inertial kernel data to obtain the forward reachability domain data; Multi-channel preservation of forward reachable domain data is calculated to obtain continuity preservation data. Based on the continuity retention data, inertial decay compensation is extrapolated to obtain compensation retention data; Based on the compensation preservation data, neighboring segment cooperative exclusion inference is performed to obtain neighboring segment competition relationship data; Based on the data on the competition relationship between adjacent segments, discontinuous crossing and bridging simulation is performed to obtain crossing and bridging data; Based on the bridging data, the continuity confidence is aggregated to obtain the segment continuity probability data; The main continuation chain is extracted based on the segment continuation probability data to obtain segment continuation relationship data.

4. The method according to claim 1, characterized in that, The specific steps for suturing broken segments across processes are as follows: Breakpoint pairing is performed based on segment continuity data to obtain breakpoint pairing data; The process traversal interpretable domain is constructed from the breakpoint pairing data to obtain joint interpretable domain data; Based on the joint interpretation domain data, a two-way evolutionary docking simulation is performed to obtain docking simulation data; Based on the docking simulation data, implicit bridging segments in the interruption zone are generated to obtain bridging segment data; Based on the bridging segment data, neighboring segment collaborative closure verification is performed to obtain segment stitching data.

5. The method according to claim 1, characterized in that, Cross-process heterogeneous coupling simulation specifically includes: By locating the process segment in the time-travel footprint data, the process segment data is obtained. The segment state transition model and continuous coupling constraint data are constructed by performing segment process data; Based on the state transition model and continuous coupling constraint data, cross-process state recursion calculations are performed to obtain cross-process simulation data.

6. The method according to claim 1, characterized in that, The reconstruction of the global heterogeneous data field specifically involves: Based on cross-process simulation data, segment state field elements are extracted to obtain segment field data. Based on the segment field metadata, heterogeneous state variables are mapped to the same domain to obtain the same domain field metadata; Based on the same domain field metadata, the segment boundaries are gradually merged to obtain boundary fusion data; Based on the boundary fusion data, cross-process layered field splicing is performed to obtain the overall field data of the continuous slab.

7. The method according to claim 6, characterized in that, The gradual fading and merging of section boundaries specifically refers to: Boundary fusion bands are constructed from the metadata of the same domain field to obtain boundary fusion band data; The difference between the two sides of the boundary is calculated based on the boundary fusion zone data to obtain the difference data between the two sides of the boundary; Distance difference transition data is obtained by calculating the distance difference transition based on the difference data on both sides of the boundary. Boundary fusion data is obtained by performing bilateral coupling fusion calculations based on distance difference transition data.

8. The method according to claim 1, characterized in that, Step S4 is as follows: The slab mapping data is obtained by meshing the overall field data of the continuous slab. Twin structure data is obtained by constructing a twin structure from the slab mapping data. The twin structure data is recursively updated to obtain the twin update data; Continuous field reconstruction is performed based on the twin update data to obtain slab twin data.

9. A cloud data-driven twin construction system for insulation board production lines, characterized in that, For executing the cloud data-driven twin construction method for insulation board production lines as described in claim 1, the cloud data-driven twin construction system for insulation board production lines includes: At least one processor; A memory, wherein the memory stores computer programs that can be executed by the processor; Communication interface, used for data interaction with production line sensing devices and cloud data platform; When the processor executes the computer program, it performs the following functions: The sensing section binding module is used to acquire continuous slab flow data; and to bind sensing sections to the continuous slab flow data to obtain sensing section data. The multidimensional time-history footprint reconstruction module is used to reconstruct the multidimensional time-history footprint based on the sensing segment data to obtain the time-history footprint data. The cross-process heterogeneous coupling simulation module is used to perform cross-process heterogeneous coupling simulation on the time-history footprint data to obtain cross-process simulation data; based on the cross-process simulation data, the global heterogeneous data field is reconstructed to obtain the overall field data of the continuous slab. The continuous twin expression module is used to continuously express twins based on the overall field data of the continuous slab, and obtain slab twin data.

Images