A method and system for real-time control of a flattening device based on edge computing

Abstract

This invention relates to the field of industrial control technology, and discloses a real-time control method and system for a flattening device based on edge computing. The method includes: transmitting the material state quantities and actuator rotation speed of the target process to a local data hub to obtain an edge process dataset; mapping process parameters to isochronous slots according to sampling timestamps to obtain a synchronous process data set; replacing the original central value with the local median value along the time direction to obtain a cleaning parameter set; correlating the cleaning parameter set with the historical parameter envelope to obtain a force-position coordination trend; determining a coordinated adjustment command based on the common trend of the current cleaning pressure value and thickness value; and issuing the command to the clamping force adjustment end and the conveying speed adjustment end to obtain an edge control loop. This invention can improve the efficiency of real-time control of a flattening device based on edge computing.

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, and in particular to a real-time control method and system for a flattening device based on edge computing. Background Technology

[0002] In existing flattening equipment control processes, the acquisition of material state parameters often relies on sampling from multiple independent sensors. The inherent discrepancies in sampling times between different sensors prevent pressure and thickness values ​​from aligning on a strictly identical time reference, thus compromising the synchronous comparability of subsequent process analyses. Furthermore, the acquired raw data contains transient sensor noise and jump values ​​caused by local anomalies on the material surface. Existing technologies typically employ simple amplitude limiting or averaging filters for processing. However, amplitude limiting filters cannot effectively eliminate continuously occurring outliers, while averaging filters obscure the true trend of process parameter changes, resulting in the cleaned parameter set failing to accurately reflect the actual state of the flattening process.

[0003] In existing technologies for coordinated control of pressure and thickness, pressure regulation and speed regulation are typically executed as two independent loops. The pressure controller outputs a clamping force adjustment amount independently based on the pressure deviation, and the speed controller outputs a conveyor speed adjustment amount independently based on the thickness deviation. There is a lack of a coordinating mechanism between the two regulation commands. When the pressure approaches the allowable limit while the thickness still has a large adjustment margin, the independent loops cannot coordinate the adjustment tendencies of the two, easily leading to situations where the speed regulation continues to increase the thickness even when the pressure is already too high, exacerbating the risk of the process exceeding the allowable limit. Furthermore, existing control loops rely on remote cloud servers for data processing and command calculation. The acquired process parameters need to undergo a long network transmission path to obtain control decisions, and transmission delays and network fluctuations directly affect the real-time performance and accuracy of the regulation commands. Summary of the Invention

[0004] This invention provides a real-time control method and system for a flattening device based on edge computing, the main purpose of which is to solve the problem of low efficiency in real-time control of flattening devices based on edge computing.

[0005] To achieve the above objectives, the present invention provides a real-time control method for a flattening device based on edge computing, comprising: The material state quantities and actuator operating speeds of the target process are transmitted to the local data hub located on the flattening equipment side to obtain the edge process dataset of the target process; The process parameters in the edge process dataset are mapped to the isochronous slots of the local data hub according to the sampling timestamps to obtain the synchronous process data group of the target process. In the synchronous process data set, the original value of the center position is replaced by the median value of the local value set along the time progression direction to obtain the cleaning parameter set of the target process; By associating the cleaning parameter set with the process envelope corresponding to the historical parameter envelope in the target process, the force-position coordination trend of the target process is obtained. The coordinated adjustment command of the target process is determined based on the common trend of the current cleaning pressure value and the current cleaning thickness value of the target process along the force-position coordination trend; The coordinated adjustment command is sent to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process.

[0006] In a preferred embodiment, the step of transmitting the material state quantities and actuator operating speeds of the target process to a local data hub located on the flattening equipment side to obtain the edge process dataset of the target process includes: Sensor collection points are arranged along the material passage path and pressure application area of ​​the target process. The sensor collection points synchronously extract the pressure value, thickness value, speed value and actuator operating speed value of the material at a uniform rhythm. The pressure value, thickness value, speed value and actuator operating speed value are bound to the corresponding sampling timestamp to obtain the real-time process parameter stream of the target process. The real-time process parameter stream is transmitted to the local data hub via the equipment field communication bus to obtain the sampling frames of the target process, and the sampling frames are collected to obtain the edge process dataset of the target process.

[0007] In a preferred embodiment, the step of mapping the process parameters in the edge process dataset to isochronous slots in the local data hub according to the sampling timestamps to obtain the synchronous process data set of the target process includes: The sampling timestamps in the edge process dataset are placed on the local clock reference line of the local data hub, and the local clock reference line is divided into isochronous slots according to a fixed time span; The sampling timestamps are assigned to the corresponding isochronous slots to obtain the synchronous process data group of the target process.

[0008] In a preferred embodiment, the step of successively replacing the original value of the center position with the median value of the local value group along the time progression direction in the synchronous process data group to obtain the cleaning parameter set of the target process includes: Arrange the same type of process parameters in the synchronous process data group according to the slot time sequence to obtain the single parameter value sequence of the target process; The fixed neighbor value group is extracted along the time progression direction of the single parameter value sequence, and the fixed neighbor value group is arranged in ascending order to obtain the ordered fixed neighbor value group of the target process; The cleaning parameter set of the target process is obtained by replacing the original parameter value at the center slot in the single parameter value sequence with the middle position value of the ordered neighbor group.

[0009] In a preferred embodiment, the step of trajectory correlation between the cleaning parameter set and the process envelope corresponding to the historical parameter envelope of the target process to obtain the force-position coordination trend of the target process includes: The force-potential envelope domain of the target process is constructed based on the pressure variation boundary and thickness target boundary of the historical parameter envelope during the target process. The real-time pressure value and real-time thickness value in the cleaning parameter set are sequentially placed into the force-position envelope domain to obtain the landing point location set of the target process. The coordinate points in the landing point location set are connected in time order to obtain the force-position change trajectory of the target process. Along the temporal progression direction of the force potential variation trajectory, the four-way boundary distances from the force potential variation trajectory to the force potential envelope are extracted sequentially to obtain the four-way boundary distance sequence of the target process. The pressure boundary direction pointed to by the pressure near-end boundary distance and the thickness boundary direction pointed to by the thickness near-end boundary distance in the four-way boundary distance sequence are respectively taken as the pressure variation tendency and the thickness variation tendency. The pressure variation tendency and the thickness variation tendency are then connected in series according to the time progression direction to obtain the pressure tendency sequence and the thickness tendency sequence of the target process. The continuous segments with adjacent equal tendency values ​​in the pressure tendency sequence and the thickness tendency sequence are respectively taken as the pressure co-directional segment and the thickness co-directional segment of the target process, thus obtaining the co-directional segment pairing set of the target process; The pressure segment and the thickness segment, which are located in the same time zone, are coupled together to obtain the pressure-thickness co-directional trend segment of the target process. The pressure-thickness co-directional trend segment is then spliced ​​together according to the time sequence progression direction to obtain the force-position co-directional trend of the target process.

[0010] In a preferred embodiment, constructing the force-potential envelope domain of the target process based on the pressure variation boundary and thickness target boundary of the historical parameter envelope of the target process includes: Extract the upper and lower boundary values ​​of the allowable pressure variation range and the upper and lower boundary values ​​of the allowable material thickness range from the historical parameter envelope to obtain the pressure envelope boundary value and thickness envelope boundary value of the target process. The intervals defined by the pressure envelope boundary value and the thickness envelope boundary value are respectively taken as the pressure axis and the thickness axis, and the pressure axis and the thickness axis are orthogonally aligned to obtain the envelope coordinate system of the target process; In the envelope coordinate system, the region enclosed by the upper pressure boundary, lower pressure boundary, upper thickness boundary, and lower thickness boundary of the historical parameters is taken as the force potential envelope domain of the target process.

[0011] In a preferred embodiment, determining the coordinated adjustment command of the target process along the force-position coordination trend based on the common trend of the current cleaning pressure value and the current cleaning thickness value of the target process includes: Based on the force-position coordination trend, the current cleaning pressure value along the pressure axis and the current cleaning thickness value along the thickness axis are intersected and positioned to obtain the current process coordinate point of the target process; The vertical distance from the current process coordinate point to the upper boundary of the target process pressure is taken as the pressure adjustment space, and the vertical distance to the lower boundary of the target process pressure is taken as the pressure adjustment space. The horizontal distance from the current process coordinate point to the upper boundary of the target process thickness is taken as the acceleration space, and the horizontal distance to the lower boundary of the target process thickness is taken as the deceleration space. The direction of the pressure boundary pointed to by the dominant distance in the pressure adjustment space and the pressure reduction space is determined as the pressure adjustment direction, and the direction of the thickness boundary pointed to by the dominant distance in the acceleration space and the deceleration space is determined as the speed adjustment direction. The dominant distance is used as the adjustment step size and adjusted along the pressure adjustment direction and the speed adjustment direction to obtain the coordinated adjustment command of the target process.

[0012] In a preferred embodiment, the formula for calculating the dominance distance includes: in, The dominant distance, To increase the pressure distance, It is a natural constant. In response to the sharpness factor, To reduce the pressure distance, This represents the total span of the pressure boundary.

[0013] In a preferred embodiment, the step of issuing the coordinated adjustment command to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process includes: The pressure adjustment command in the coordinated adjustment command is transmitted to the clamping force adjustment end, driving the clamping force adjustment end to adjust the clamping degree of the material, thereby obtaining the feedback pressure value of the target process; The speed adjustment command in the coordinated adjustment command is transmitted to the conveying speed adjustment end, driving the conveying speed adjustment end to adjust the conveying speed of the material, and obtaining the feedback speed value of the target process; The feedback pressure value and the feedback speed value are incorporated into the real-time process parameter acquisition stream of the target process to obtain the edge control loop of the target process.

[0014] To address the aforementioned problems, the present invention also provides a real-time control system for a flattening device based on edge computing, the system comprising: The data acquisition module transmits the material state quantities and actuator operating speed of the target process to the local data hub located on the flattening equipment side to obtain the edge process dataset of the target process; The data synchronization module maps the process parameters in the edge process dataset to the isochronous slots of the local data hub according to the sampling timestamp, thereby obtaining the synchronized process data group of the target process. The cleaning parameter module replaces the original value of the center position with the median value of the local value group along the time progression direction in the synchronous process data group to obtain the cleaning parameter set of the target process. The force-position coordination trend module correlates the cleaning parameter set with the process envelope corresponding to the historical parameter envelope in the target process to obtain the force-position coordination trend of the target process. The coordinated adjustment instruction module determines the coordinated adjustment instruction of the target process according to the common trend of the current cleaning pressure value and the current cleaning thickness value of the target process along the force-position coordination trend. The edge control loop module sends the coordinated adjustment command to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. By transmitting material state parameters and actuator operating speed to a local data hub located on the flattening equipment side to form an edge process dataset, and then mapping it to isochronous slots according to the sampling timestamp, a synchronous process data group is obtained. All process parameters are strictly aligned on the same time baseline, eliminating the data synchronization problem caused by sampling time deviation. A cleaning method is adopted to successively replace the original value of the center position with the median value of the local value group along the time progression direction. The original value of each center slot is replaced by the median value of its neighborhood window. This effectively eliminates isolated jump points caused by instantaneous sensor noise and local anomalies on the material surface, while completely preserving the true change trend of pressure and thickness during the flattening process. By trajectory correlation between the cleaning parameter set and the process envelope corresponding to the historical parameter envelope, the force-position coordination trend is obtained. The part of the continuous change tendency of pressure and thickness that overlaps in time is identified as the pressure-thickness coordination trend segment, thus accurately describing the joint change law of force and position throughout the flattening process.

[0016] 2. Based on the force-position coordination trend, the current cleaning pressure value and the current cleaning thickness value are intersected and located in the envelope coordinate system. The vertical and lateral distances from the current process coordinate point to the upper and lower pressure boundaries, the upper and lower thickness boundaries are calculated respectively, obtaining the pressure adjustment space, pressure reduction space, speed increase space, and speed decrease space. Then, a weighted fusion calculation is used to obtain the dominant distance and the direction of the pressure boundary and thickness boundary it points to. The dominant distance is used as the adjustment step size to generate coordinated adjustment commands along the pressure adjustment direction and the speed adjustment direction. The size of each adjustment step size is adaptively correlated with the size of the currently available space, realizing a precise match between the adjustment range and the adjustment requirements. The pressure adjustment command is transmitted to the clamping force adjustment end, and the speed adjustment command is transmitted to the conveying speed adjustment end. The feedback pressure value and feedback speed value are incorporated into the real-time process parameter acquisition stream, forming a complete edge control loop. The entire control cycle is completed locally on the equipment side, significantly reducing control latency and improving the real-time performance and accuracy of the adjustment commands. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a real-time control method for a flattening device based on edge computing, provided in an embodiment of the present invention. Figure 2 This is a functional block diagram of a real-time control system for a flattening device based on edge computing, provided in an embodiment of the present invention. The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides a real-time control method for a flattening device based on edge computing. The execution entity of this real-time control method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the real-time control method for a flattening device based on edge computing can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating a real-time control method for a flattening device based on edge computing, according to an embodiment of the present invention. In this embodiment, the real-time control method for a flattening device based on edge computing includes: In this embodiment of the invention, the step of transmitting the material state quantities and actuator operating speed of the target process to a local data hub located on the flattening equipment side to obtain the edge process dataset of the target process is specifically used for: Sensor collection points are arranged along the material passage path and pressure application area of ​​the target process. The sensor collection points synchronously extract the pressure value, thickness value, speed value and actuator operating speed value of the material at a uniform rhythm. The pressure value, thickness value, speed value and actuator operating speed value are bound to the corresponding sampling timestamp to obtain the real-time process parameter stream of the target process. The real-time process parameter stream is transmitted to the local data hub via the equipment field communication bus to obtain the sampling frames of the target process, and the sampling frames are collected to obtain the edge process dataset of the target process.

[0021] Specifically, in the material passage path and pressure application area of ​​the target process, pressure sensors and thickness sensors are installed on the inlet and outlet sides of the flattening roller respectively along the material movement direction. At the same time, a speed sensor is installed on the rotating shaft of the actuator and a speed sensor is installed on the conveyor belt drive wheel. All these sensors generate a unified synchronous sampling cycle through the same hardware clock source.

[0022] Specifically, each record in the real-time process parameter stream is sent one by one to the local data hub located on the flattening equipment side via the equipment field communication bus in the order of its sampling time. After receiving each record, the local data hub encapsulates it into an independent sampling frame. This sampling frame retains all the information in the original record, namely the sampling timestamp, pressure value, thickness value, speed value, and rotation speed value.

[0023] Furthermore, at the instant of each cycle, all sensors simultaneously collect their corresponding physical quantity values: the pressure sensor measures the material pressure value at the current moment, the thickness sensor measures the material thickness value at the current moment, the speed sensor measures the material conveying speed value at the current moment, and the rotation speed sensor measures the actuator operating speed value at the current moment. Subsequently, the acquisition system binds these four values ​​with the timestamp of the sampling to form a complete record containing the sampling time, pressure value, thickness value, speed value, and rotation speed value. By repeating each sampling cycle, a continuous sequence of records is obtained, which is the real-time process parameter stream of the target process.

[0024] Furthermore, all values ​​collected within the same sampling cycle are encapsulated in the same sampling frame. The local data hub stores all received sampling frames in local memory without omission, according to the order of the sampling timestamps. When all sampling frames from the start of monitoring to the current time are collected, the overall set of these sampling frames constitutes the edge process dataset of the target process.

[0025] In summary, by arranging sensor acquisition points along the material passage path and in the pressure application area, and synchronously extracting pressure, thickness, speed, and actuator rotation speed values ​​at a unified pace, and binding sampling timestamps, all process parameters originate from the material state at the same moment. This eliminates the data synchronization problem caused by differences in acquisition time between different sensors, thus providing an accurate time alignment basis for subsequent synchronized process data sets. This ensures that each set of data in the real-time process parameter stream truly reflects the flattening process state at the same instant.

[0026] In summary, real-time process parameter streams are transmitted to the local data hub via the equipment's field communication bus and encapsulated into sampling frames. All sampling frames are then aggregated to form an edge process dataset. This allows the originally scattered sensor data to be centrally stored and managed on the equipment side, avoiding the transmission delays and network fluctuations caused by uploading raw data to a remote cloud. At the same time, the sampling frame format retains the complete timestamp and process parameter correspondence, providing a complete and directly usable local data foundation for subsequent isochronous slot mapping and cleaning parameter set construction.

[0027] In this embodiment of the invention, the step of mapping the process parameters in the edge process dataset to the isochronous slots of the local data hub according to the sampling timestamps to obtain the synchronous process data group of the target process is specifically used for: The sampling timestamps in the edge process dataset are placed on the local clock reference line of the local data hub, and the local clock reference line is divided into isochronous slots according to a fixed time span; The sampling timestamps are assigned to the corresponding isochronous slots to obtain the synchronous process data group of the target process.

[0028] Specifically, the local data hub extracts the sampling timestamp carried by each sampling frame in the edge process data set, and places the sampling timestamp as a time coordinate point on the local clock reference line maintained inside the local data hub. The local clock reference line is a continuous straight line with real physical time as its axis, and its zero point is the start time of the local data hub. Each sampling timestamp is precisely projected onto a unique position on this straight line according to the time difference between it and the zero point.

[0029] Specifically, the local data hub traverses all sampling frames in the edge process dataset, extracts the sampling timestamp for each sampling frame, and compares the value corresponding to the sampling timestamp with the start and end times of each isochronous slot.

[0030] Furthermore, the local data hub then divides this local clock baseline from zero along the direction of time increase into adjacent and non-overlapping time intervals with a fixed and uniform time span. Each such time interval is called an isochronous slot, and these isochronous slots are arranged sequentially to cover the entire time axis. Each isochronous slot has a defined start time and end time.

[0031] Furthermore, the unique isochronous slot is found where the start time is less than or equal to the sampling timestamp and the end time is greater than the sampling timestamp. The sampling frame is determined to belong to this isochronous slot. The local data hub arranges the process parameters in all sampling frames belonging to the same isochronous slot according to the order of the sampling timestamps to form the process parameter set of that slot. The process parameter sets corresponding to all isochronous slots are combined according to the time order of the slots to form the synchronous process data group of the target process.

[0032] In summary, placing the sampling timestamps in the edge process dataset on the local clock reference of the local data hub and dividing them into isochronous slots with fixed time spans ensures that the slight offsets that may exist due to sampling times from different sensors are uniformly aligned to the same discretized time grid. This eliminates the problem of inconsistent time references caused by uneven distribution of sampling timestamps and provides an equally spaced time reference frame for subsequent synchronization of process data sets, thereby ensuring that the process parameters in each slot are truly synchronous and comparable.

[0033] In summary, by assigning the sampling timestamps to the corresponding isochronous slots, synchronous process data groups are obtained. This allows the original asynchronously arriving sampling frames to be regrouped according to unified isochronous slots. Each slot gathers all process parameters within the same time window, thereby achieving strict alignment of pressure, thickness, speed, and rotation speed values ​​in the time dimension. This eliminates misalignment interference caused by data transmission jitter and slight offsets in sampling time, providing a time-synchronized and structurally sound data foundation for subsequent cleaning parameter sets.

[0034] In this embodiment of the invention, when the original value of the center position is successively replaced with the median value of the local value group along the time progression direction in the synchronous process data group to obtain the cleaning parameter set of the target process, it is specifically used for: Arrange the same type of process parameters in the synchronous process data group according to the slot time sequence to obtain the single parameter value sequence of the target process; The fixed neighbor value group is extracted along the time progression direction of the single parameter value sequence, and the fixed neighbor value group is arranged in ascending order to obtain the ordered fixed neighbor value group of the target process; The cleaning parameter set of the target process is obtained by replacing the original parameter value at the center slot in the single parameter value sequence with the middle position value of the ordered neighbor group.

[0035] Specifically, the local data hub extracts a certain category of process parameters, such as pressure values, from the synchronous process data group, and retrieves all values ​​of that category of process parameters in each slot according to the time sequence of the isochronous slots.

[0036] Specifically, the local data hub moves a fixed-length time window sequentially along the time progression direction in the single-parameter value sequence, starting from the beginning of the sequence. This time window covers a consecutive odd number of slots, with the central slot located in the exact middle of the window.

[0037] Specifically, the local data hub extracts the value located in the middle position from the ordered neighbor value group. This value is the median value of the neighbor value group. The local data hub finds the center slot in the original time window corresponding to the neighbor value group and replaces the original parameter value in the center slot in the single parameter value sequence with this median value.

[0038] Furthermore, if a slot contains the same type of process parameter from multiple sampling frames, these values ​​are arranged sequentially according to the sampling timestamps. If a slot does not contain the same type of process parameter, the slot is left empty in the sequence. After processing all slots sequentially, a long sequence is obtained by arranging the slot timestamps. This sequence is the single parameter value sequence of the target process.

[0039] Furthermore, each time the window stays, the values ​​of all the slots covered within the window are extracted to form a local value group corresponding to the window range. This local value group is called the fixed neighbor value group. The local data hub rearranges all the values ​​in this fixed neighbor value group in ascending order to obtain a new sequence. This new sequence is called the ordered fixed neighbor value group.

[0040] Furthermore, after the replacement is completed, the local data hub moves the time window forward by one slot and repeats the operations of intercepting adjacent value groups, sorting in ascending order, extracting the median value, and replacing the original value of the center slot until all slots in the single parameter value sequence that can form a complete window have been traversed. Finally, the new values ​​on all the replaced center slots are arranged according to the original slot time sequence to form the cleaning parameter set of the target process.

[0041] In summary, arranging similar process parameters in the synchronous process data group according to the slot time sequence yields a single parameter value sequence. This organizes the same physical quantity, which was originally scattered in various equal-time slots, into a complete time sequence, facilitating subsequent local window processing along the time direction. At the same time, it preserves the original time sequence position of each slot, providing an accurate order basis for the extraction of adjacent value groups.

[0042] In summary, by extracting adjacent value groups along the time progression direction of the single parameter value sequence and arranging them in ascending order, an ordered adjacent value group is obtained. This allows the values ​​within each local window to be rearranged in order of size, thereby accurately identifying the median value in the middle order within the window. This sorting process eliminates the random fluctuation order of the original values ​​and provides a correct basis for selecting the middle order for replacing the original value of the center slot with the median value.

[0043] In summary, by replacing the original parameter value at the center slot in the single parameter value sequence with the median value of the ordered neighboring value group, the cleaning parameter set is obtained. This allows the original pressure value, thickness value, or rotation speed value at each center slot to be replaced by the median value of its neighborhood window. This effectively eliminates isolated jump points caused by instantaneous sensor noise or local anomalies on the material surface, while preserving the overall trend of process parameters. This provides a smooth and accurate parameter set that reflects the process state for subsequent force-position synergy trend analysis.

[0044] In this embodiment of the invention, the step of associating the cleaning parameter set with the process envelope corresponding to the historical parameter envelope of the target process to obtain the force-position coordination trend of the target process is specifically used for: The force-potential envelope domain of the target process is constructed based on the pressure variation boundary and thickness target boundary of the historical parameter envelope during the target process. The real-time pressure value and real-time thickness value in the cleaning parameter set are sequentially placed into the force-position envelope domain to obtain the landing point location set of the target process. The coordinate points in the landing point location set are connected in time order to obtain the force-position change trajectory of the target process. Along the temporal progression direction of the force potential variation trajectory, the four-way boundary distances from the force potential variation trajectory to the force potential envelope are extracted sequentially to obtain the four-way boundary distance sequence of the target process. The pressure boundary direction pointed to by the pressure near-end boundary distance and the thickness boundary direction pointed to by the thickness near-end boundary distance in the four-way boundary distance sequence are respectively taken as the pressure variation tendency and the thickness variation tendency. The pressure variation tendency and the thickness variation tendency are then connected in series according to the time progression direction to obtain the pressure tendency sequence and the thickness tendency sequence of the target process. The continuous segments with adjacent equal tendency values ​​in the pressure tendency sequence and the thickness tendency sequence are respectively taken as the pressure co-directional segment and the thickness co-directional segment of the target process, thus obtaining the co-directional segment pairing set of the target process; The pressure segment and the thickness segment, which are located in the same time zone, are coupled together to obtain the pressure-thickness co-directional trend segment of the target process. The pressure-thickness co-directional trend segment is then spliced ​​together according to the time sequence progression direction to obtain the force-position co-directional trend of the target process.

[0045] Specifically, the local data hub extracts the upper and lower bound values ​​of the allowable pressure variation from the historical parameter envelope as the pressure variation boundary, and at the same time extracts the upper and lower bound values ​​of the allowable material thickness as the thickness target boundary. Then, a planar coordinate system is established with the pressure value as the vertical axis and the thickness value as the horizontal axis.

[0046] Specifically, the local data hub extracts the real-time pressure and real-time thickness values ​​from each set of data in the order of sampling time from the set of cleaning parameters. The real-time pressure value is used as the vertical axis and the real-time thickness value is used as the horizontal axis to mark a unique coordinate point in the plane coordinate system where the force-position envelope domain is located. This coordinate point is recorded as the landing point at that moment.

[0047] Specifically, starting from the initial endpoint of the force potential variation trajectory, the local data hub extracts each coordinate point on the force potential variation trajectory point by point along the time progression direction. For each coordinate point, the vertical distance from the point to the horizontal line where the upper boundary of the force potential envelope domain, i.e., the upper boundary of the pressure, is located is calculated as the upper boundary distance; the vertical distance from the point to the horizontal line where the lower boundary of the force potential envelope domain, i.e., the lower boundary of the pressure, is located is calculated as the lower boundary distance; the horizontal distance from the point to the vertical line where the left boundary of the force potential envelope domain, i.e., the lower boundary of the thickness, is located is calculated as the left boundary distance; and the horizontal distance from the point to the vertical line where the right boundary of the force potential envelope domain, i.e., the upper boundary of the thickness, is located is calculated as the right boundary distance.

[0048] Specifically, the local data hub traverses each quadruple in the four-way boundary distance sequence, takes out the upper boundary distance and the lower boundary distance and compares them. The boundary distance with the smaller value is called the pressure proximal boundary distance and the boundary direction pointed to by the boundary distance is recorded. That is, if the upper boundary distance is smaller, the pressure proximal boundary distance points to the upper boundary of pressure, and if the lower boundary distance is smaller, it points to the lower boundary of pressure. This direction is taken as the pressure change tendency at that moment.

[0049] Specifically, the local data hub scans backward from the starting position of the pressure tendency sequence to find all segments with identical tendency values ​​at consecutive positions. Each such segment is called a pressure-in-the-direction segment. Similarly, it scans backward from the starting position of the thickness tendency sequence to find all segments with identical tendency values ​​at consecutive positions. Each such segment is called a thickness-in-the-direction segment.

[0050] Specifically, the local data hub traverses each pair of pressure-oriented segments and thickness-oriented segments in the same-direction segment pairing set, and determines whether the time intervals covered by these two segments overlap, i.e., the same time zone. If there is an overlap, the pressure variation tendency and thickness variation tendency corresponding to the overlapping time interval are coupled together to form a pressure-thickness co-directional segment.

[0051] Furthermore, the upper and lower pressure boundary lines are drawn as two horizontal lines parallel to the horizontal axis, and the upper and lower thickness boundary lines are drawn as two vertical lines parallel to the vertical axis. These four lines intersect to form a rectangular region, which is the force potential envelope of the target process.

[0052] Furthermore, this operation is repeated for the cleaning parameters at all times to obtain a series of coordinate points arranged in chronological order. The set of these coordinate points is called the landing point location set. The local data hub connects the adjacent coordinate points in the landing point location set in chronological order with straight line segments to form a broken line that continuously passes through each landing point from the earliest time to the latest time. This broken line is called the force-potential variation trace.

[0053] Furthermore, these four distance values ​​are combined into a quadruple in the order of top, bottom, left, and right. Each coordinate point on the force potential change trajectory is processed in turn, and the quadruple corresponding to each coordinate point is arranged in chronological order. The resulting sequence is called the four-way boundary distance sequence.

[0054] Furthermore, the left-hand boundary distance and the right-hand boundary distance are compared simultaneously. The boundary distance with the smaller value is called the near-end boundary distance of the thickness, and the boundary direction pointed to by this boundary distance is recorded. That is, if the left-hand boundary distance is smaller, it points to the lower boundary of the thickness, and if the right-hand boundary distance is smaller, it points to the upper boundary of the thickness. This direction is taken as the thickness variation tendency at that moment. The sequence formed by arranging these pressure variation tendencies in chronological order is called the pressure tendency sequence, and the sequence formed by arranging these thickness variation tendencies in chronological order is called the thickness tendency sequence.

[0055] Furthermore, all pressure-in-the-direction segments and all thickness-in-the-direction segments are paired according to the order in which they appear in their respective sequences, that is, the first pressure-in-the-direction segment is paired with the first thickness-in-the-direction segment, the second pressure-in-the-direction segment is paired with the second thickness-in-the-direction segment, and so on. The set of all pairing results is called the pairing set of segments.

[0056] Furthermore, this segment contains coordinated information that the pressure and thickness both tend towards a certain boundary during this time period. By piecing together all the obtained pressure-thickness coordinated trend segments in chronological order, the overall trend description formed is the force-position coordinated trend of the target process.

[0057] In summary, a force-position envelope domain is constructed based on the pressure variation boundary and thickness target boundary of the historical parameter envelope. This provides a clear allowable operating range for subsequent real-time process data as a reference benchmark. This envelope domain unifies the boundary constraints of the two physical quantities, pressure and thickness, into a two-dimensional coordinate system, providing a fixed spatial framework for the insertion of the landing point location set and trajectory association. This ensures that the force-position collaborative trend analysis is always carried out within the safe boundaries allowed by the equipment.

[0058] In summary, by sequentially placing the real-time pressure and thickness values ​​from the cleaning parameter set into the force-position envelope domain to obtain the set of landing point locations, and connecting them in chronological order to obtain the force-position change trajectory, the originally isolated and scattered post-cleaning data forms a continuous time evolution trajectory on the pressure-thickness plane, intuitively showing the movement path of the process state within the envelope domain. At the same time, the temporal information of the cleaning parameter set is transformed into spatial geometric information, providing a visualized geometric basis for subsequent extraction of four-way boundary distances and judgment of change trends.

[0059] In summary, by sequentially extracting the four-way boundary distances along the temporal progression of the force potential variation trajectory, a four-way boundary distance sequence is obtained. This quantifies the distances of the pressure and thickness values ​​at each moment relative to the four boundaries (upper, lower, left, and right) of the force potential envelope into four distinct values. This transforms the complex spatial relationships into a one-dimensional distance sequence, facilitating subsequent point-by-point analysis of the degree to which the process state approaches each boundary. At the same time, the four-way boundary distance sequence retains the complete temporal order, providing direct numerical basis for calculating the pressure variation tendency and thickness variation tendency.

[0060] In summary, the pressure boundary direction pointed to by the pressure near-end boundary distance in the four-way boundary distance sequence is taken as the pressure variation trend, and the thickness boundary direction pointed to by the thickness near-end boundary distance is taken as the thickness variation trend. These are then concatenated in time sequence to obtain the pressure trend sequence and the thickness trend sequence. This simplifies the pressure change trend and the thickness change trend at each moment into a direction value pointing to a specific boundary, eliminating redundant information from the original distance values. At the same time, the two trend sequences are aligned with the same time axis, providing synchronous discretized directional data for subsequent identification of co-directional segments and pressure-thickness co-analysis.

[0061] In summary, by taking consecutive segments with equal adjacent trend values ​​in the pressure tendency sequence and thickness tendency sequence as pressure-oriented segments and thickness-oriented segments respectively, a pairing set of oriented segments is obtained. This merges the originally discrete and potentially frequently changing direction values ​​into continuous time periods with stable trends. Each oriented segment represents the process state continuously tending towards the same pressure boundary or the same thickness boundary over a period of time. The pairing set of oriented segments corresponds the pressure-oriented segments and the thickness-oriented segments one-to-one in chronological order, providing well-matched time interval pairs for correlation and coupling.

[0062] In summary, by pairing and coupling pressure and thickness segments that are concentrated in the same time zone, a pressure-thickness co-directional trend segment is obtained. This segment is then spliced ​​together according to the temporal progression to obtain the force-position co-directional trend. This allows the overlapping parts of the continuous change tendencies of pressure and thickness in time to be identified. Each pressure-thickness co-directional trend segment represents the co-directional behavior of pressure and thickness simultaneously and continuously tending towards their respective boundaries. The force-position co-directional trend formed by splicing all segments in chronological order fully describes the joint change law of force and position during the entire flattening process, providing a direct trend basis for the subsequent generation of co-directional adjustment commands.

[0063] In this embodiment of the invention, when constructing the force potential envelope domain of the target process based on the pressure variation boundary and thickness target boundary of the historical parameter envelope of the target process, it is specifically used for: Extract the upper and lower boundary values ​​of the allowable pressure variation range and the upper and lower boundary values ​​of the allowable material thickness range from the historical parameter envelope to obtain the pressure envelope boundary value and thickness envelope boundary value of the target process. The intervals defined by the pressure envelope boundary value and the thickness envelope boundary value are respectively taken as the pressure axis and the thickness axis, and the pressure axis and the thickness axis are orthogonally aligned to obtain the envelope coordinate system of the target process; In the envelope coordinate system, the region enclosed by the upper pressure boundary, lower pressure boundary, upper thickness boundary, and lower thickness boundary of the historical parameters is taken as the force potential envelope domain of the target process.

[0064] Specifically, the local data hub reads the upper limit of the allowable pressure variation from the historical parameter envelope as the upper pressure boundary value, and reads the lower limit of the allowable pressure variation as the lower pressure boundary value. The upper and lower pressure boundary values ​​are collectively referred to as the pressure envelope boundary values.

[0065] Specifically, the local data hub uses the entire continuous interval from the lower boundary value to the upper boundary value of pressure, defined by the pressure envelope boundary value, as the scale range of the pressure axis, and the entire continuous interval from the lower boundary value to the upper boundary value of thickness, defined by the thickness envelope boundary value, as the scale range of the thickness axis. Then, the pressure axis is used as the vertical axis and the thickness axis is used as the horizontal axis.

[0066] Specifically, the local data hub draws a horizontal line in the envelope coordinate system that is parallel to the thickness axis and passes through the location of the upper pressure boundary value on the pressure axis as the upper pressure boundary line, and draws a horizontal line that is parallel to the thickness axis and passes through the location of the lower pressure boundary value on the pressure axis as the lower pressure boundary line.

[0067] Furthermore, the upper limit value of the allowable variation of material thickness is read from the historical parameter envelope as the upper boundary value of thickness, and the lower limit value of the allowable variation of material thickness is read as the lower boundary value of thickness. The upper boundary value and the lower boundary value of thickness are collectively referred to as the thickness envelope boundary value.

[0068] Furthermore, the two axes are placed perpendicularly to each other at their respective zero points, so that the value of the pressure axis increases vertically upwards and the value of the thickness axis increases horizontally to the right. The coordinate system formed after orthogonal alignment is called the envelope coordinate system.

[0069] Furthermore, a vertical line parallel to the pressure axis and passing through the location of the upper thickness boundary value on the thickness axis is drawn as the upper thickness boundary line, and a vertical line parallel to the pressure axis and passing through the location of the lower thickness boundary value on the thickness axis is drawn as the lower thickness boundary line. The rectangular closed region enclosed by the intersection of these four lines is called the force potential envelope.

[0070] In summary, by extracting the upper and lower boundary values ​​of the allowable range of pressure variation and the allowable range of material thickness from the historical parameter envelope, the pressure envelope boundary values ​​and thickness envelope boundary values ​​are obtained. This ensures that the boundary values ​​on which the subsequent envelope coordinate system is constructed are entirely derived from the historical allowable range during normal equipment operation, avoiding the arbitrariness of manually setting boundaries. At the same time, by extracting the boundary values ​​of the two different physical quantities, pressure and thickness, independently, clear interval endpoints in two directions are provided for subsequent orthogonal alignment.

[0071] In summary, by orthogonally aligning the intervals defined by the pressure envelope boundary values ​​and the thickness envelope boundary values ​​as the pressure axis and the thickness axis respectively, an envelope coordinate system is obtained. This transforms the allowable ranges of the two dimensions of pressure and thickness into two coordinate axes with a perpendicular relationship. The scale range on each axis corresponds exactly to the legal value range of the corresponding physical quantity. The orthogonal alignment ensures that the pressure change and the thickness change are independent and perpendicular to each other in the coordinate system, providing a standard two-dimensional spatial positioning framework for accurately placing real-time process coordinate points into the force-position envelope domain.

[0072] In summary, in the envelope coordinate system, the region enclosed by the upper and lower pressure boundaries, upper and lower thickness boundaries of the historical parameters is taken as the force-potential envelope domain. This makes the envelope domain a clearly defined, closed and continuous rectangular area. Each coordinate point within the area represents a set of process states that simultaneously meet the allowable pressure and thickness ranges. Outside the area, it represents exceeding the allowable range. This provides an intuitive spatial basis for subsequent judgment of whether the real-time pressure and thickness values ​​are within the safe range and their distance from each boundary.

[0073] In this embodiment of the invention, when determining the coordinated adjustment command of the target process based on the common trend of the current cleaning pressure value and the current cleaning thickness value along the force-position coordination trend, it is specifically used for: Based on the force-position coordination trend, the current cleaning pressure value along the pressure axis and the current cleaning thickness value along the thickness axis are intersected and positioned to obtain the current process coordinate point of the target process; The vertical distance from the current process coordinate point to the upper boundary of the target process pressure is taken as the pressure adjustment space, and the vertical distance to the lower boundary of the target process pressure is taken as the pressure adjustment space. The horizontal distance from the current process coordinate point to the upper boundary of the target process thickness is taken as the acceleration space, and the horizontal distance to the lower boundary of the target process thickness is taken as the deceleration space. The direction of the pressure boundary pointed to by the dominant distance in the pressure adjustment space and the pressure reduction space is determined as the pressure adjustment direction, and the direction of the thickness boundary pointed to by the dominant distance in the acceleration space and the deceleration space is determined as the speed adjustment direction. The dominant distance is used as the adjustment step size and adjusted along the pressure adjustment direction and the speed adjustment direction to obtain the coordinated adjustment command of the target process.

[0074] Specifically, the local data hub extracts the real-time pressure value at the current moment from the cleaning parameter set as the current cleaning pressure value, and extracts the real-time thickness value at the current moment as the current cleaning thickness value, based on the pressure change tendency and thickness change tendency corresponding to the current moment in the force-position coordination trend.

[0075] Specifically, the local data hub draws a vertical line segment between the current process coordinate point and the upper pressure boundary of the force envelope domain, and measures the length of this line segment as the pressure adjustment space. It also draws a vertical line segment between the current process coordinate point and the lower pressure boundary, and measures the length of this line segment as the pressure adjustment space. Similarly, it draws a horizontal line segment between the current process coordinate point and the upper thickness boundary, and measures the length of this line segment as the acceleration space. Finally, it draws a horizontal line segment between the current process coordinate point and the lower thickness boundary.

[0076] Specifically, the local data hub inputs the values ​​of the upward and downward pressure spaces, as well as the total span of the pressure boundary, into a weighted fusion calculation along with a pre-set response sharpness coefficient. This calculation assigns weights related to the response sharpness coefficient to the upward and downward pressure spaces respectively. The weight allocation gives higher weights to the pressure spaces with larger values. Then, the two weighted pressure space values ​​are averaged, and the final output value is called the dominance distance.

[0077] Specifically, the local data hub uses the value of the dominant distance calculated in the pressure regulation direction as the pressure regulation step size and the value of the dominant distance calculated in the speed regulation direction as the speed regulation step size, and then generates a coordinated regulation command that includes pressure regulation command and speed regulation command.

[0078] Furthermore, the current cleaning pressure value is projected onto the pressure axis along the pressure axis direction of the envelope coordinate system, and the current cleaning thickness value is projected onto the thickness axis along the thickness axis direction. Straight lines perpendicular to their respective axes are drawn from the projection points on the pressure axis and the projection points on the thickness axis. The intersection of these two perpendicular lines is a unique coordinate point in the envelope coordinate system, which is called the current process coordinate point.

[0079] Furthermore, the length of the line segment is measured as the deceleration space, where the upward pressure space and the downward pressure space both represent the difference between the current pressure value and the upper and lower boundaries of the allowable pressure range, and the speed-up space and the speed-down space both represent the difference between the current thickness value and the upper and lower boundaries of the allowable thickness range.

[0080] Furthermore, the value of this dominant distance is closer to the larger value in the upward pressure adjustment space and the downward pressure adjustment space. Therefore, based on whether the dominant distance is closer to the upward pressure adjustment space or the downward pressure adjustment space, we can determine the direction of the pressure boundary it points to and determine this direction as the pressure adjustment direction. Then, we perform the same weighted fusion calculation on the acceleration space and the deceleration space to obtain another dominant distance. Based on whether this dominant distance is closer to the acceleration space or the deceleration space, we can determine the direction of the thickness boundary it points to and determine this direction as the speed adjustment direction.

[0081] Furthermore, the pressure regulation command specifies the regulation amount corresponding to the dominant distance moved along the pressure regulation direction, and the speed regulation command specifies the regulation amount corresponding to the dominant distance moved along the speed regulation direction. The complete command formed by combining these two commands is called the coordinated regulation command.

[0082] In summary, based on the force-position synergy trend, the current cleaning pressure value along the pressure axis and the current cleaning thickness value along the thickness axis are respectively used to locate the current process coordinate point. This ensures that the pressure and thickness values ​​after cleaning are accurately mapped to a unique coordinate point in the envelope coordinate system. This coordinate point also reflects the specific position of the current pressure and thickness combination state in the force-position envelope domain, providing a unique spatial reference point for subsequent calculations of the pressure adjustment space, pressure reduction space, speed increase space, and speed decrease space.

[0083] In summary, by defining the vertical distance from the current process coordinate point to the upper pressure boundary as the pressure adjustment space, the vertical distance to the lower pressure boundary as the pressure adjustment space, the lateral distance to the upper thickness boundary as the speed-up space, and the lateral distance to the lower thickness boundary as the speed-down space, the remaining adjustment margin of the current process state relative to the four allowable boundaries is quantified into four distinct spatial values. The pressure adjustment space and pressure reduction space indicate the available range for increasing or decreasing the pressure, while the speed-up space and speed-down space indicate the available range for increasing or decreasing the thickness. This provides direct numerical basis for subsequently determining the dominant distance and adjustment direction.

[0084] In summary, the direction of the pressure boundary pointed to by the dominant distance in the pressure adjustment space and the pressure reduction space is determined as the pressure adjustment direction, and the direction of the thickness boundary pointed to by the dominant distance in the speed increase space and the speed decrease space is determined as the speed adjustment direction. This means that the selection of the adjustment direction no longer depends on manual experience or fixed thresholds, but automatically biases towards the side with larger available space after weighted fusion calculation based on the available space in the two directions. This ensures that the adjustment command always moves towards the direction with a safer process state and more sufficient margin, avoiding the risk of exacerbating the approach to the boundary due to incorrect direction selection.

[0085] In summary, by using the dominant distance as the adjustment step size along both the pressure and speed adjustment directions, a coordinated adjustment command is obtained. This allows the adjustment step size to adaptively correlate with the available space. When the available space in a certain direction is larger, the dominant distance is closer to the distance value of that space, resulting in a larger step size. When the available space in two directions is similar, the dominant distance takes the middle value, resulting in a moderate step size. This achieves a precise match between the adjustment range and the adjustment requirements. At the same time, the pressure adjustment and speed adjustment are integrated into a single command, providing a unified operational basis for the coordinated action of the subsequent clamping force adjustment end and the conveying speed adjustment end.

[0086] In this embodiment of the invention, the formula for calculating the dominance distance is specifically used for: in, The dominant distance, To increase the pressure distance, It is a natural constant. In response to the sharpness factor, To reduce the pressure distance, This represents the total span of the pressure boundary.

[0087] Specifically, the upward pressure distance is derived from the measured vertical distance from the current process coordinate point to the upper pressure boundary, the downward pressure distance is derived from the measured vertical distance from the current process coordinate point to the lower pressure boundary, the total pressure boundary span is derived from the total distance between the upper and lower pressure boundaries, the natural constant is a fixed mathematical constant, and the response sharpness coefficient is a fixed value pre-set according to the equipment response characteristics. These values ​​are fed into the weighted fusion calculation.

[0088] Furthermore, the function of this calculation formula is to calculate a weighted average between the upward and downward pressure distances. This weighted average is not a simple arithmetic average, but rather assigns a higher weight to the larger pressure distance, making the calculation result closer to the larger pressure distance. At the same time, the response sharpness coefficient controls the sensitivity of this bias. The larger the response sharpness coefficient, the more the calculation result is biased towards the larger pressure distance. The final output advantage distance is used to determine the pressure adjustment direction and as the adjustment step size.

[0089] In general, when the upward pressure distance is much greater than the downward pressure distance, the dominant distance is close to the upward pressure distance. At this time, the pressure adjustment direction points to the upper boundary and the step size is large. When the downward pressure distance is much greater than the upward pressure distance, the dominant distance is close to the downward pressure distance. At this time, the pressure adjustment direction points to the lower boundary and the step size is large. When the upward pressure distance and the downward pressure distance are equal, the dominant distance is equal to the common value of the two. At this time, either direction can be selected. When the response sharpness coefficient is larger, the dominant distance is more sensitive to the difference between the two pressure distances, and vice versa.

[0090] In this embodiment of the invention, when the coordinated adjustment command is sent to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process, it is specifically used for: The pressure adjustment command in the coordinated adjustment command is transmitted to the clamping force adjustment end, driving the clamping force adjustment end to adjust the clamping degree of the material, thereby obtaining the feedback pressure value of the target process; The speed adjustment command in the coordinated adjustment command is transmitted to the conveying speed adjustment end, driving the conveying speed adjustment end to adjust the conveying speed of the material, and obtaining the feedback speed value of the target process; The feedback pressure value and the feedback speed value are incorporated into the real-time process parameter acquisition stream of the target process to obtain the edge control loop of the target process.

[0091] Specifically, the local data hub extracts the pressure adjustment command from the coordinated adjustment command and transmits the pressure adjustment command to the clamping force adjustment end of the flattening equipment through the equipment field communication bus. After receiving the pressure adjustment command, the clamping force adjustment end parses out the pressure adjustment direction and pressure adjustment step size contained therein.

[0092] Specifically, the local data hub extracts the speed adjustment command from the coordinated adjustment command and transmits the speed adjustment command to the conveying speed adjustment end of the flattening equipment through the equipment field communication bus. After receiving the speed adjustment command, the conveying speed adjustment end parses out the speed adjustment direction and speed adjustment step size contained therein.

[0093] Specifically, the local data hub binds the received feedback pressure and feedback velocity values ​​to the corresponding sampling timestamps according to the time they are generated. Then, it inserts these two new sets of values ​​into the real-time process parameter acquisition stream that is currently being continuously acquired, so that the real-time process parameter acquisition stream is updated to the latest data sequence containing feedback pressure and feedback velocity values ​​based on the original pressure, thickness, velocity, and rotation speed values.

[0094] Furthermore, the actuator then drives itself to change the clamping force output on the flattening roller according to the direction and step size, so that the clamping degree of the flattening roller increases or decreases accordingly. After the clamping force adjustment end completes the adjustment action, it measures the actual pressure value acting on the material in real time through the built-in pressure sensor, and transmits the actual pressure value as the feedback pressure value back to the local data hub.

[0095] Furthermore, the frequency converter driving the conveyor motor changes the running speed of the conveyor belt according to the direction and step size, so that the material conveying speed is increased or decreased accordingly. After the conveying speed adjustment end completes the adjustment action, it measures the actual running speed value of the current conveyor belt in real time through the built-in speed sensor, and transmits the actual speed value as the feedback speed value back to the local data hub.

[0096] Furthermore, at this point, a complete closed loop is formed from data acquisition, data synchronization, data cleaning, trend analysis, instruction generation to instruction execution and feedback. This closed loop is the edge control loop of the target process.

[0097] In summary, the pressure regulation command in the coordinated regulation command is transmitted to the clamping force regulation end, which drives the clamping force regulation end to adjust the clamping degree of the material to obtain the feedback pressure value. This allows the pressure regulation command to be implemented from the calculation level to the actual action of the actuator. At the same time, a closed-loop feedback is formed by measuring the actual pressure value after adjustment in real time, which provides direct verification data of the current pressure regulation effect for the subsequent edge control loop.

[0098] In summary, the speed adjustment command in the coordinated adjustment command is transmitted to the conveying speed adjustment end, which drives the conveying speed adjustment end to adjust the conveying speed of the material to obtain a feedback speed value. This allows the speed adjustment command to be converted into an actual change in the rotational speed of the conveying motor. The actual speed value after adjustment is measured in real time and transmitted back to the local data hub, ensuring the observability and verifiability of the speed control process.

[0099] In summary, by incorporating the feedback pressure and speed values ​​into the real-time process parameter acquisition stream to obtain the edge control loop, the adjustment results and the original acquired data are merged in the same data stream, completing the entire closed loop from data acquisition, synchronization, cleaning, trend analysis, instruction generation to execution feedback. The local data hub can independently complete the entire control cycle without relying on a remote server, significantly reducing control latency and improving the real-time performance and reliability of the system.

[0100] Compared with the prior art, the present invention has the following beneficial effects: 1. By transmitting material state parameters and actuator operating speed to a local data hub located on the flattening equipment side to form an edge process dataset, and then mapping it to isochronous slots according to the sampling timestamp, a synchronous process data group is obtained. All process parameters are strictly aligned on the same time baseline, eliminating the data synchronization problem caused by sampling time deviation. A cleaning method is adopted to successively replace the original value of the center position with the median value of the local value group along the time progression direction. The original value of each center slot is replaced by the median value of its neighborhood window. This effectively eliminates isolated jump points caused by instantaneous sensor noise and local anomalies on the material surface, while completely preserving the true change trend of pressure and thickness during the flattening process. By trajectory correlation between the cleaning parameter set and the process envelope corresponding to the historical parameter envelope, the force-position coordination trend is obtained. The part of the continuous change tendency of pressure and thickness that overlaps in time is identified as the pressure-thickness coordination trend segment, thus accurately describing the joint change law of force and position throughout the flattening process.

[0101] 2. Based on the force-position coordination trend, the current cleaning pressure value and the current cleaning thickness value are intersected and located in the envelope coordinate system. The vertical and lateral distances from the current process coordinate point to the upper and lower pressure boundaries, the upper and lower thickness boundaries are calculated respectively, obtaining the pressure adjustment space, pressure reduction space, speed increase space, and speed decrease space. Then, a weighted fusion calculation is used to obtain the dominant distance and the direction of the pressure boundary and thickness boundary it points to. The dominant distance is used as the adjustment step size to generate coordinated adjustment commands along the pressure adjustment direction and the speed adjustment direction. The size of each adjustment step size is adaptively correlated with the size of the currently available space, realizing a precise match between the adjustment range and the adjustment requirements. The pressure adjustment command is transmitted to the clamping force adjustment end, and the speed adjustment command is transmitted to the conveying speed adjustment end. The feedback pressure value and feedback speed value are incorporated into the real-time process parameter acquisition stream, forming a complete edge control loop. The entire control cycle is completed locally on the equipment side, significantly reducing control latency and improving the real-time performance and accuracy of the adjustment commands.

[0102] like Figure 2 The diagram shown is a functional block diagram of a real-time control system for a flattening device based on edge computing, provided in an embodiment of the present invention.

[0103] The edge computing-based real-time control system 100 for a flattening device described in this invention can be installed in an electronic device. Depending on the functions implemented, the edge computing-based real-time control system 100 may include a data acquisition module 101, a data synchronization module 102, a cleaning parameter module 103, a force-position coordination trend module 104, a coordination adjustment command module 105, and an edge control loop module 106. The module described in this invention can also be called a unit, referring to a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, stored in the memory of the electronic device.

[0104] In this embodiment, the functions of each module / unit are as follows: The data acquisition module transmits the material state quantities and actuator operating speed of the target process to the local data hub located on the flattening equipment side to obtain the edge process dataset of the target process; The data synchronization module maps the process parameters in the edge process dataset to the isochronous slots of the local data hub according to the sampling timestamp, thereby obtaining the synchronized process data group of the target process. The cleaning parameter module replaces the original value of the center position with the median value of the local value group along the time progression direction in the synchronous process data group to obtain the cleaning parameter set of the target process. The force-position coordination trend module correlates the cleaning parameter set with the process envelope corresponding to the historical parameter envelope in the target process to obtain the force-position coordination trend of the target process. The coordinated adjustment instruction module determines the coordinated adjustment instruction of the target process according to the common trend of the current cleaning pressure value and the current cleaning thickness value of the target process along the force-position coordination trend. The edge control loop module sends the coordinated adjustment command to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process.

[0105] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0106] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0107] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0108] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0109] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0110] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A real-time control method for a flattening device based on edge computing, characterized in that, The method includes: The material state quantities and actuator operating speeds of the target process are transmitted to the local data hub located on the flattening equipment side to obtain the edge process dataset of the target process; The process parameters in the edge process dataset are mapped to the isochronous slots of the local data hub according to the sampling timestamps to obtain the synchronous process data group of the target process. In the synchronous process data set, the original value of the center position is replaced by the median value of the local value set along the time progression direction to obtain the cleaning parameter set of the target process; By associating the cleaning parameter set with the process envelope corresponding to the historical parameter envelope in the target process, the force-position coordination trend of the target process is obtained. The coordinated adjustment command of the target process is determined based on the common trend of the current cleaning pressure value and the current cleaning thickness value of the target process along the force-position coordination trend; The coordinated adjustment command is sent to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process.

2. The real-time control method for a flattening device based on edge computing as described in claim 1, characterized in that, The process of transmitting the material state parameters and actuator operating speed of the target process to a local data hub located on the flattening equipment side to obtain the edge process dataset of the target process includes: Sensor collection points are arranged along the material passage path and pressure application area of ​​the target process. The sensor collection points synchronously extract the pressure value, thickness value, speed value and actuator operating speed value of the material at a uniform rhythm. The pressure value, thickness value, speed value and actuator operating speed value are bound to the corresponding sampling timestamp to obtain the real-time process parameter stream of the target process. The real-time process parameter stream is transmitted to the local data hub via the equipment field communication bus to obtain the sampling frames of the target process, and the sampling frames are collected to obtain the edge process dataset of the target process.

3. The real-time control method for a flattening device based on edge computing as described in claim 1, characterized in that, The step of mapping the process parameters in the edge process dataset to the isochronous slots of the local data hub according to the sampling timestamp to obtain the synchronous process data group of the target process includes: The sampling timestamps in the edge process dataset are placed on the local clock reference line of the local data hub, and the local clock reference line is divided into isochronous slots according to a fixed time span; The sampling timestamps are assigned to the corresponding isochronous slots to obtain the synchronous process data group of the target process.

4. The real-time control method for a flattening device based on edge computing as described in claim 1, characterized in that, The process involves successively replacing the original value of the center position with the median value of the local value group along the time progression direction in the synchronous process data group to obtain the cleaning parameter set of the target process, including: Arrange the same type of process parameters in the synchronous process data group according to the slot time sequence to obtain the single parameter value sequence of the target process; The fixed neighbor value group is extracted along the time progression direction of the single parameter value sequence, and the fixed neighbor value group is arranged in ascending order to obtain the ordered fixed neighbor value group of the target process; The cleaning parameter set of the target process is obtained by replacing the original parameter value at the center slot in the single parameter value sequence with the middle position value of the ordered neighbor group.

5. The real-time control method for a flattening device based on edge computing as described in claim 1, characterized in that, The step of associating the cleaning parameter set with the process envelope corresponding to the historical parameter envelope in the target process to obtain the force-position coordination trend of the target process includes: The force-potential envelope domain of the target process is constructed based on the pressure variation boundary and thickness target boundary of the historical parameter envelope during the target process. The real-time pressure value and real-time thickness value in the cleaning parameter set are sequentially placed into the force-position envelope domain to obtain the landing point location set of the target process. The coordinate points in the landing point location set are connected in time order to obtain the force-position change trajectory of the target process. Along the temporal progression direction of the force potential variation trajectory, the four-way boundary distances from the force potential variation trajectory to the force potential envelope are extracted sequentially to obtain the four-way boundary distance sequence of the target process. The pressure boundary direction pointed to by the pressure near-end boundary distance and the thickness boundary direction pointed to by the thickness near-end boundary distance in the four-way boundary distance sequence are respectively taken as the pressure variation tendency and the thickness variation tendency. The pressure variation tendency and the thickness variation tendency are then connected in series according to the time progression direction to obtain the pressure tendency sequence and the thickness tendency sequence of the target process. The continuous segments with adjacent equal tendency values ​​in the pressure tendency sequence and the thickness tendency sequence are respectively taken as the pressure co-directional segment and the thickness co-directional segment of the target process, thus obtaining the co-directional segment pairing set of the target process; The pressure segment and the thickness segment, which are located in the same time zone, are coupled together to obtain the pressure-thickness co-directional trend segment of the target process. The pressure-thickness co-directional trend segment is then spliced ​​together according to the time sequence progression direction to obtain the force-position co-directional trend of the target process.

6. The real-time control method for a flattening device based on edge computing as described in claim 5, characterized in that, The construction of the force-potential envelope domain of the target process based on the pressure variation boundary and thickness target boundary of the historical parameter envelope of the target process includes: Extract the upper and lower boundary values ​​of the allowable pressure variation range and the upper and lower boundary values ​​of the allowable material thickness range from the historical parameter envelope to obtain the pressure envelope boundary value and thickness envelope boundary value of the target process. The intervals defined by the pressure envelope boundary value and the thickness envelope boundary value are respectively taken as the pressure axis and the thickness axis, and the pressure axis and the thickness axis are orthogonally aligned to obtain the envelope coordinate system of the target process; In the envelope coordinate system, the region enclosed by the upper pressure boundary, lower pressure boundary, upper thickness boundary, and lower thickness boundary of the historical parameters is taken as the force potential envelope domain of the target process.

7. The real-time control method for a flattening device based on edge computing as described in claim 1, characterized in that, The determination of the coordinated adjustment command for the target process based on the common trend of the current cleaning pressure value and the current cleaning thickness value along the force-position coordination trend includes: Based on the force-position coordination trend, the current cleaning pressure value along the pressure axis and the current cleaning thickness value along the thickness axis are intersected and positioned to obtain the current process coordinate point of the target process; The vertical distance from the current process coordinate point to the upper boundary of the target process pressure is taken as the pressure adjustment space, and the vertical distance to the lower boundary of the target process pressure is taken as the pressure adjustment space. The horizontal distance from the current process coordinate point to the upper boundary of the target process thickness is taken as the acceleration space, and the horizontal distance to the lower boundary of the target process thickness is taken as the deceleration space. The direction of the pressure boundary pointed to by the dominant distance in the pressure adjustment space and the pressure reduction space is determined as the pressure adjustment direction, and the direction of the thickness boundary pointed to by the dominant distance in the acceleration space and the deceleration space is determined as the speed adjustment direction. The dominant distance is used as the adjustment step size and adjusted along the pressure adjustment direction and the speed adjustment direction to obtain the coordinated adjustment command of the target process.

8. The real-time control method for a flattening device based on edge computing as described in claim 7, characterized in that, The formula for calculating the dominant distance includes: in, The dominant distance, To increase the pressure distance, It is a natural constant. In response to the sharpness factor, To reduce the pressure distance, This represents the total span of the pressure boundary.

9. The real-time control method for a flattening device based on edge computing as described in claim 1, characterized in that, The step of issuing the coordinated adjustment command to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process includes: The pressure adjustment command in the coordinated adjustment command is transmitted to the clamping force adjustment end, driving the clamping force adjustment end to adjust the clamping degree of the material, thereby obtaining the feedback pressure value of the target process; The speed adjustment command in the coordinated adjustment command is transmitted to the conveying speed adjustment end, driving the conveying speed adjustment end to adjust the conveying speed of the material, and obtaining the feedback speed value of the target process; The feedback pressure value and the feedback speed value are incorporated into the real-time process parameter acquisition stream of the target process to obtain the edge control loop of the target process.

10. A real-time control system for a flattening device based on edge computing, used to implement the real-time control method for a flattening device based on edge computing as described in any one of claims 1-9, characterized in that, The system includes: The data acquisition module transmits the material state quantities and actuator operating speed of the target process to the local data hub located on the flattening equipment side to obtain the edge process dataset of the target process; The data synchronization module maps the process parameters in the edge process dataset to the isochronous slots of the local data hub according to the sampling timestamp, thereby obtaining the synchronized process data group of the target process. The cleaning parameter module replaces the original value of the center position with the median value of the local value group along the time progression direction in the synchronous process data group to obtain the cleaning parameter set of the target process. The force-position coordination trend module correlates the cleaning parameter set with the process envelope corresponding to the historical parameter envelope in the target process to obtain the force-position coordination trend of the target process. The coordinated adjustment instruction module determines the coordinated adjustment instruction of the target process according to the common trend of the current cleaning pressure value and the current cleaning thickness value of the target process along the force-position coordination trend. The edge control loop module sends the coordinated adjustment command to the clamping force adjustment end and the conveying speed adjustment end of the target process to obtain the edge control loop of the target process.

Images