Edge computing-based intelligent manufacturing production line data communication scheduling system

Abstract

This invention discloses a data communication scheduling system for intelligent manufacturing production lines based on edge computing, belonging to the field of industrial IoT edge computing data scheduling technology. The system synchronously acquires the real-time load current discrete sequence of the target device node and the real-time signal-to-noise ratio sequence of the main communication path, extracting the second-order derivative zero-crossing discrete component and the first-order attenuation discrete component of the envelope. A cross-correlation algorithm is used to perform coherence analysis to derive a cross-domain healthy coupling index. Using this index as the independent variable, a weighted mapping of the real-time packet loss rate is performed through an exponential gain function to generate a virtual impedance factor. It determines whether the index exceeds a threshold to trigger backup gateway context pre-synchronization and calculates the positive offset. Finally, a normalization function is used to calculate the traffic allocation weight and configure path scheduling labels to achieve dynamic scheduling of multi-path traffic splitting actuators. This solution achieves a priori prediction of industrial communication link quality and refined multi-path traffic splitting scheduling.

Description

Technical Field

[0001] This invention relates to the field of edge computing data scheduling technology in the Industrial Internet of Things (IIoT) field, and more specifically, this application relates to a smart manufacturing production line data communication scheduling system based on edge computing. Background Technology

[0002] In the digital transformation of smart manufacturing production lines, the stability of communication on the industrial site directly affects production safety and efficiency. Currently, most industrial production lines use wireless communication technology in conjunction with edge computing nodes to transmit control commands and monitoring data.

[0003] However, intelligent manufacturing production lines are scenarios where high-intensity electromagnetic interference and complex mechanical operations coexist. During operation, the transient fluctuations in the physical load of target equipment such as high-power motors and cutting machine tools are often accompanied by severe pulsed electromagnetic noise.

[0004] Traditional communication scheduling techniques primarily rely on feedback mechanisms within the communication layer itself, such as triggering backup link switching by statistically analyzing historical packet loss rates or monitoring real-time signal-to-noise ratio fluctuations. This approach has significant drawbacks: First, changes in communication parameters often lag behind fluctuations in the physical environment. When the packet loss rate reaches a threshold, the data link may already be substantially blocked, leading to irreversible loss or delay of critical control commands. Second, existing systems often lack smooth transition mechanisms during the switching process, exhibiting a context reset blind spot at the moment of primary / backup path switching, making it difficult to cope with complex situations where transient load surges and network degradation occur simultaneously.

[0005] In the highly dynamic electromagnetic environment of intelligent manufacturing, existing communication scheduling systems lack the ability to predict the correlation between physical load and communication links. This results in significant latency blind spots and switching jitter issues in path scheduling when faced with interference caused by sudden heavy loads on equipment.

[0006] To address the aforementioned issues, there is an urgent need in this field for an intelligent communication scheduling scheme that can deeply correlate the physical load characteristics of devices with the quality of communication links, and perform forward-looking preprocessing and smooth traffic distribution before the link collapses. Summary of the Invention

[0007] To address the aforementioned technical issues, this technical solution provides a data communication and scheduling system for intelligent manufacturing production lines based on edge computing. The solution resolves the problems mentioned in the background section.

[0008] In a first aspect, embodiments of this application provide a smart manufacturing production line data communication scheduling system based on edge computing, comprising: a component acquisition module: used to simultaneously acquire the real-time load current discrete sequence of the target device node and the real-time signal-to-noise ratio sequence of the main communication path within a preset observation window, and extract the second derivative zero-crossing discrete component of the real-time load current discrete sequence and the envelope first-order attenuation discrete component of the real-time signal-to-noise ratio sequence; a coupling index processing module: used to perform coherence analysis on the second derivative zero-crossing discrete component and the envelope first-order attenuation discrete component with the same sampling frequency using a cross-correlation algorithm to obtain a cross-domain health coupling index; and a factor processing module: used to use the cross-domain health coupling index as an independent variable and a preset exponential gain function to process the main communication path. The system performs a weighted mapping of the real-time packet loss rate of the path to generate a virtual impedance factor. The offset processing module determines whether the virtual impedance factor exceeds a preset impedance threshold. If it does, it sends a context pre-synchronization command to the backup gateway and calculates the positive offset of the virtual impedance factor exceeding the preset impedance threshold. The communication scheduling module inputs the positive offset into the Sigmoid normalization function to calculate the traffic allocation weight and configures a path scheduling label containing path selection variables for the data stream to be sent. The multi-path splitter executes the data stream path mapping transmission between the main communication path and the backup path based on the path selection variables in the path scheduling label, and outputs the standardized data scheduling result for the current scheduling period.

[0009] Secondly, this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the aforementioned edge computing-based intelligent manufacturing production line data communication scheduling system.

[0010] One or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages:

[0011] 1. Existing technologies often rely solely on passive switching based on communication layer parameters, resulting in significant hysteresis. This solution extracts the zero-crossing discrete components of the second derivative of the real-time load current discrete sequence to identify electromagnetic interference characteristics generated by transient load changes in equipment at the physical source. Furthermore, it performs coherence analysis with the first-order attenuation discrete components of the envelope of the real-time signal-to-noise ratio sequence. The resulting cross-domain health coupling index can reflect the deterioration trend of the communication environment in advance, achieving deep cross-domain prediction of physical health status and communication quality.

[0012] 2. By generating a virtual impedance factor, a context pre-synchronization command is sent to the backup gateway in advance using a positive offset before the main communication path completely collapses. This "predictive" processing avoids the logical reset delay that exists when switching paths in traditional solutions, ensures the continuity of transmission of critical data such as control commands in complex industrial electromagnetic environments, and constructs a preventive pre-synchronization buffer mechanism.

[0013] 3. By using the positive offset to calculate the traffic allocation weight through a normalization function, and in conjunction with the path selection variable in the path scheduling label, a smooth transition from single-path transmission to multi-path proportional traffic splitting is achieved. This solves the resource waste problem caused by directly performing a full switch in the early stages of primary path deterioration. At the same time, the intervention of backup paths offsets the performance fluctuations of the main communication path, significantly reducing the end-to-end average jitter of data and providing smooth and adaptive dynamic traffic scheduling capabilities. Attached Figure Description

[0014] Figure 1 A schematic diagram of the structure of a smart manufacturing production line data communication and scheduling system based on edge computing provided in an embodiment of this application;

[0015] Figure 2 A schematic diagram of the logical flow of a smart manufacturing production line data communication scheduling system based on edge computing, provided in an embodiment of this application. Detailed Implementation

[0016] This application provides an edge computing-based intelligent manufacturing production line data communication scheduling system, which solves the technical problem in the prior art that, in the highly dynamic electromagnetic environment of intelligent manufacturing, due to the lack of correlation prediction capability between physical load and communication link, path scheduling has significant time delay blind spots and switching jitter when facing interference caused by sudden heavy loads of equipment.

[0017] This solution considers that the root cause of communication environment degradation in intelligent production lines is often the fluctuation of mechanical operating loads on equipment. It simultaneously acquires the real-time discrete sequence of load current at equipment nodes and the real-time signal-to-noise ratio (SNR) sequence of the main communication path. Since simple current values ​​are insufficient to capture transient changes, the solution extracts the zero-crossing discrete component of the second derivative to pinpoint the moment of current abrupt change and correlates it with the first-order attenuation discrete component of the envelope in the SNR sequence, which reflects the channel attenuation trend. This cross-domain health coupling index, derived through a cross-correlation algorithm, essentially reveals the endogenous relationship between physical vibration loads and electromagnetic environment degradation, elevating the dimension for judging communication quality from a single link layer to a joint "physical-communication" domain.

[0018] After establishing the perception depth, this scheme quantifies this cross-domain risk by constructing a virtual impedance factor. Using an exponential gain function, the virtual impedance factor exhibits a non-linear growth as the cross-domain health coupling exponent increases; this growth trend is far more sensitive than changes in the actual packet loss rate. By determining the relationship between this factor and a preset impedance threshold, this scheme can preemptively calculate the positive offset. The deeper significance of this offset lies in measuring the degree of spillover of current communication risks. It not only triggers context pre-synchronization of the backup path, providing environmental preparation for the "soft landing" of the data flow, but also transforms it into precise traffic allocation weights through a normalization function.

[0019] Ultimately, this solution configures path scheduling labels to instruct the multi-path load balancing executor to perform fine-grained load balancing based on path selection variables. Under normal risk fluctuations, the system achieves load balancing through proportional mapping; however, when extreme conflicts such as heavy load surges and interference deterioration are detected, the system can switch from proportional load balancing mode to full-redundant concurrent mode. By modifying the path selection variable to a full-path broadcast variable and executing zero-latency concurrent transmission on both paths, this solution achieves deterministic protection of the core data flow during peak physical conflict moments. The entire derivation logic starts with physical-level feature extraction, proceeds through cross-domain mapping and risk quantification, and ultimately closes the loop with multi-mode adaptive scheduling, fundamentally solving the scheduling instability problem of existing systems in dynamic production environments.

[0020] To better understand the above technical solutions, the following will provide a detailed explanation of the technical solutions in conjunction with the accompanying drawings and specific implementation methods.

[0021] This embodiment provides an edge computing-based intelligent manufacturing production line data communication scheduling system. Its core application scenario is an intelligent manufacturing production line where high-intensity electromagnetic interference and complex mechanical operations coexist, specifically, an automotive body-in-white welding production line simultaneously operating multiple high-power servo motors, frequency converters, and welding robots. The system aims to solve the latency blind spots and switching jitter problems in existing technologies caused by a lack of predictive capabilities regarding the correlation between physical load and communication links. By deeply correlating the physical load characteristics of equipment with the quality of communication links, the system performs forward-looking preprocessing and smooth traffic distribution before link failure, achieving prior prediction of industrial communication link quality and refined multi-path traffic distribution scheduling.

[0022] like Figure 1 The diagram shown is a schematic of the data communication scheduling system for intelligent manufacturing production lines based on edge computing provided in this application embodiment; it includes: a component acquisition module: used to synchronously acquire the real-time load current discrete sequence of the target device node and the real-time signal-to-noise ratio sequence of the main communication path within a preset observation window.

[0023] In this embodiment, the length of the preset observation window is... This is not a fixed value; its value is set based on the typical load fluctuation cycle of the target equipment. By collecting current data from the target equipment, such as a spot welding robot, during a complete work cycle including start-up, acceleration, stable welding, deceleration, and stopping, the autocorrelation function is analyzed to determine the lag time corresponding to the energy concentration region as the observation window length. The typical engineering value is set to 500 milliseconds, with a range of 100 to 1000 milliseconds. Real-time load current discrete sequence. A high-frequency current sensor deployed on the power supply line of the target device is used to sample the frequency. Data is collected at Hz, meaning 1000 discrete sampling points are acquired per second, indexed. ,in Real-time signal-to-noise ratio sequence of the main communication path The current is periodically reported by 5GURLLC industrial modules deployed on edge computing nodes. The reporting period is strictly synchronized with the current sampling period. The milliseconds ensure that the two sequences are homologous on the time axis, providing a precise alignment basis for subsequent coherence analysis.

[0024] Extract the second derivative zero-crossing discrete component of the real-time load current discrete sequence and the first-order decay discrete component of the envelope of the real-time signal-to-noise ratio sequence.

[0025] The cross-correlation algorithm is used to perform coherence analysis on the second-order derivative zero-crossing discrete components and the first-order decaying discrete components of the envelope with the same sampling frequency, and the cross-domain health coupling index is obtained.

[0026] Specifically, this embodiment calculates the cross-domain health coupling index using the following formula. :

[0027] ,in, Indicates the cross-domain health coupling index. The second derivative zero-crossing discrete component, For the first-order decay discrete component of the envelope, The first-order decay discrete component of the envelope after discrete displacement translation. The total number of sampling points within the preset observation window, i.e. , This is a discrete sampling time index, with a value range of [value range missing]. , These are discrete displacement parameters, with a value range of [value range missing]. This parameter is used when calculating the cross-correlation function. Compared to The iterative variables for translation optimization. It is a function that takes the maximum peak value in the sequence of cross-correlation results.

[0028] At different time offsets Below, calculate the load impact signal of the computing device. With communication link attenuation signal The sum of the dot products of the two is maximized when they are optimally aligned on the time axis; this is the cross-domain health coupling index. The higher this peak value, the more temporally coupled the transient impacts in the physical domain are with the link deterioration in the communication domain.

[0029] Using the cross-domain health coupling index as the independent variable, a virtual impedance factor is generated by weighting the real-time packet loss rate of the main communication path through a preset exponential gain function.

[0030] Real-time packet loss rate Indexed by the receiving edge node of the main communication path at each discrete sampling time. The corresponding real-time statistics within a 1-millisecond short time slot are calculated as follows: ,in This represents the total number of data packets sent by the transmitter within this time slot. This represents the number of data packets not received by the receiving end. This embodiment generates the virtual impedance factor using the following formula. :

[0031] ,in, Represents the virtual impedance factor. Index of the main communication path at the current discrete sampling time Real-time packet loss rate As a cross-domain health coupling index, It is a natural exponential function. This is the preset sensitivity correction factor. This is used to control the gain strength of physical domain perturbations on the virtual impedance factor, and its typical value ranges from 0.5 to 2.0. This embodiment uses engineering experience values ​​under the default environment. When the system needs to be more sensitive to physical load shocks, i.e., trigger pre-synchronization earlier, the setting can be increased. Up to 2.0; if the electromagnetic environment of the production line itself is excellent and you do not want it to be too sensitive, you can lower it to 0.5.

[0032] Determine if the virtual impedance factor is greater than the preset impedance threshold. If it is, send a context pre-synchronization command to the backup gateway and calculate the positive offset of the virtual impedance factor from the preset impedance threshold.

[0033] Preset impedance threshold The setup follows the offline calibration procedure: In the experimental environment, a standard interference pulse sequence is applied to the target device, gradually increasing the interference intensity, while simultaneously monitoring the actual transmission status of the main communication path. Record the virtual impedance factor value at the point where continuous packet loss begins on the main path, for example, when the packet loss rate exceeds 5% for three consecutive sampling time slots, or in the last stable state before link reconnection. .

[0034] To ensure the system has sufficient reaction time before the link completely fails, a preset impedance threshold is set. In this embodiment, after calibration, the settings are... When judging At this point, the system determines that the communication risk has overflowed the security boundary. Immediately, a context pre-synchronization command is sent to the backup gateway. This command is sent in the form of a User Datagram Protocol (UDP) packet and contains key state information of the current communication session, including: the Transmission Control Protocol (TCP) connection 5-tuple, the Real-Time Transport Protocol (RTP) sequence number base, and the next transmission sequence number for the Encryption Security Protocol (ECP). Simultaneously, the positive offset is calculated. This offset is a dimensionless positive real number, and its magnitude directly quantifies the degree of current risk spillover.

[0035] The positive offset is input into the Sigmoid normalization function to calculate the traffic allocation weight. Based on this, a path scheduling label containing path selection variables is configured for the data stream to be sent. The multi-path traffic splitter executes the data stream path mapping transmission between the main communication path and the backup path based on the path selection variables in the path scheduling label. The standardized data scheduling result of the current scheduling period is output.

[0036] This embodiment calculates the traffic allocation weight using the following formula. :

[0037] ,in, Assign weights to traffic, with a value range of 100%. , This is a positive offset. The preset weight transition smoothing coefficient is used, and , It is a natural exponential function.

[0038] Preset weight transition smoothing coefficient The steepness of the Sigmoid curve transitioning from 0 to 1 is controlled through a grid search experiment in this embodiment. This value can be set in the positive offset direction. When the weight changes from 0 to 1.5, the allocation weight is adjusted. The voltage level smoothly increases from approximately 0.1 to approximately 0.9, achieving a gradual diversion. This represents the proportion of data packets that should be allocated to the backup path within the current scheduling cycle. Subsequently, a path scheduling label is configured for each data packet in the data stream to be sent. This label is an 8-bit bitmap variable named Path Selection Var.

[0039] When Path Selection Var=0x00, it means only the main path is used; when Path Selection Var=0x01, it means traffic is distributed according to weight ratio; and when Path Selection Var=0x02, it means the backup path is forced to be used. In this step, according to... The value is used to determine the tag value for each data packet using a round-robin algorithm. For example, if... Then, out of every 10 data packets, 3 data packets are randomly selected and their Path Selection Var is set to 0x02, while the remaining 7 are set to 0x00.

[0040] Finally, the multi-path routing executor. For example, a data plane development kit module running in the kernel mode of an edge computing node reads the Path Selection Var tag of each data packet and sends the data packet to the corresponding network interface card queue according to the tag value. The primary path corresponds to eth0, and the backup path corresponds to eth1, completing the data flow path mapping transmission based on traffic allocation weights. At the end of this scheduling cycle, a standardized data scheduling result is output. This result is a structured log record, including: timestamp, virtual impedance factor for this cycle. Assigning weights And the actual number of data packets transmitted on the primary / backup path.

[0041] Figure 2 This is a schematic diagram of the logical flow of a smart manufacturing production line data communication scheduling system based on edge computing, provided in an embodiment of this application. Through the aforementioned core basic scheme, this embodiment establishes a cross-domain mapping and risk quantification framework from load current in the physical domain to signal-to-noise ratio and packet loss rate in the communication domain. This solves the core defect of existing technologies that cannot proactively perceive the deterioration trend of communication links, achieving a priori prediction of link collapse risks and smooth traffic pre-sharing based on risk spillover, significantly reducing the end-to-end average jitter of the data stream.

[0042] Furthermore, in concurrent conflict scenarios, the continuous proportional traffic splitting mechanism, lacking joint perception of "impact speed" and "impact intensity," cannot distinguish between slow deterioration and instantaneous impact, resulting in out-of-order data packets or instantaneous overload of backup paths.

[0043] For example, in extreme concurrent scenarios such as when the injection molding machine closes the mold or when the stamping line starts, the motor current suddenly rises by more than 3 times the rated value within a few milliseconds, and at the same time the signal-to-noise ratio of the main communication path drops by more than 20dB.

[0044] To address the scheduling failures and data flow disorder issues caused by the conventional proportional traffic splitting strategy's inability to quantify the impact intensity and degradation rate in extreme concurrency scenarios where physical shocks and network degradation occur simultaneously, concurrency conflict determination and scheduling reconfiguration are performed before data flow path mapping and transmission between the main communication path and backup path based on traffic allocation weights.

[0045] The preset observation window is divided into several micro windows, and a synchronous mapping relationship between the scheduling cycle of the virtual impedance factor and the micro windows is established, wherein each scheduling cycle contains at least one micro window.

[0046] Set a preset observation window. For example, a window with a length of 500 milliseconds, and divide it evenly into... Each micro-window has a length of [number] micro-windows. Milliseconds. Establish synchronization mapping: Let the scheduling cycle index... With micro window index satisfy ,in The number of microwindows contained in each scheduling cycle. This embodiment sets... That is, a scheduling cycle is 100 milliseconds, which contains 10 consecutive micro windows.

[0047] The synchronous mapping relationship between the scheduling period of the virtual impedance factor and the micro-window of the discrete sequence of real-time load current is obtained.

[0048] This embodiment achieves the mapping through timestamp alignment. For scheduling cycles... Its covered time range is .

[0049] All micro windows within this interval Mapping to the same scheduling period and extracting discrete subsets of real-time load current sequences within these micro-windows. .

[0050] Calculate the average value of the moving variance of the discrete sequence of real-time load current for all micro windows within the current scheduling period, and normalize it to obtain the normalized average moving variance, which is then recorded as the transient impact characteristic.

[0051] In each micro window Within this context, the sliding variance is calculated for the discrete sequence of real-time load current. Let the sliding window length be... The sampling point. For the first sampling point within the window. There are data points, and the local variance is . The maximum value of all local variances within the microscopic window is taken as the impact intensity of that window. Calculate the current scheduling period. The average value of the impact intensity of all microscopic windows within: Finally, normalization is performed: ,in The reference variance corresponding to the rated current fluctuation of the target equipment can be obtained by calibrating the data from the stable no-load operation of the equipment. This is denoted as transient impact characteristic. Its range is A typical value exceeding 1.5 indicates a strong impact.

[0052] Calculate the discrete absolute value of the virtual impedance factor in the current scheduling period relative to the previous scheduling period, and denote it as the degradation gradient feature characterizing the rate of communication degradation.

[0053] This embodiment calculates the degradation gradient characteristics. :

[0054] ,in, The virtual impedance factor for the current scheduling period. This is the value from the previous scheduling period. This value directly reflects the rate at which communication risks deteriorate over a period of time, such as 100 milliseconds.

[0055] The product of the degraded gradient feature and the preset time delay conversion coefficient is obtained to get the dynamic compensation time delay value.

[0056] This embodiment obtains the preset time delay conversion coefficient. The gradient in milliseconds per unit is calibrated as follows: In the experimental setting, at different rates... To mitigate the degradation of the communication link, the optimal waiting time from packet transmission to being suspended in the out-of-order compensation queue and finally sent is measured, and a linear coefficient is fitted. Dynamic delay value. The calculation is as follows:

[0057] , The unit is milliseconds, and its physical meaning is: the amount of extra time that data packets on the backup path need to wait compared to the main path in order to offset the time difference between the paths.

[0058] The transient impact characteristics and degradation gradient characteristics are weighted and summed to obtain the comprehensive conflict evaluation index, and it is determined whether the comprehensive conflict evaluation index is greater than the preset conflict threshold.

[0059] This embodiment calculates the comprehensive conflict assessment index using the following formula. :

[0060] ,in, and To pre-determine the weights, this embodiment uses the analytic hierarchy process (AHP) to determine them. , This emphasizes the dominant role of physical impact. A preset conflict threshold is set. Field testing confirmed that running standard interference test cases, gradually increasing the physical impact intensity and communication degradation rate, allowed observation of unacceptable out-of-order data stream behavior. Examples include out-of-order distances exceeding 64 packets or packet loss. The value is taken as the 95th percentile as the threshold. This embodiment calibrates... . judge Whether it is valid or not.

[0061] If the value is not greater than the specified value, the multi-path traffic splitter remains in out-of-order compensation mode, executing sequentially after the "calculate traffic allocation weight" step but before the multi-path traffic splitter actually sends data packets. It is a conditionally triggered branch processing logic.

[0062] If the value is greater than the specified value, the multi-path shunt executor will be forcibly switched to concurrent circuit breaker mode. This is also a condition-triggered branch processing logic that will completely cover the proportional shunt behavior set in the core basic solution.

[0063] By introducing the above technical solution, a comprehensive conflict evaluation index weighted by transient impact characteristics and degradation gradient characteristics is introduced to accurately identify the extreme scenario of physical impact and network degradation occurring concurrently. The scheduling mode is adaptively switched according to the conflict level, which solves the problem of out-of-order and overload that may be caused by conventional proportional traffic splitting in extreme concurrency scenarios, and realizes the robust switching of scheduling strategy.

[0064] Furthermore, if the production line is in a steady state but there is moderate interference in the communication environment, such as when multiple frequency converters are simultaneously performing asynchronous speed regulation, the signal-to-noise ratio of the main path will slowly decrease, and there will be a stable propagation delay difference of 5-15 milliseconds between the backup path and the main path. In this case, under the multi-path proportional split mode, there will be a problem of disordered data packet reception order caused by the inherent propagation delay difference between the main and backup paths and minor jitter.

[0065] The processing logic of this scheme is executed sequentially after the data packet is marked as "to the backup path" by the path scheduling label, but before it is submitted to the underlying network driver for physical transmission. It constitutes an insert-type buffer queue located between the virtual device layer and the physical device layer.

[0066] Intercept packets containing path scheduling tags pointing to backup paths and suspend them by writing them to the local packet sending cache queue.

[0067] The packet filter module in the multipath splitter intercepts all packets where the path selection variable Path SelectionVar is equal to 0x02, pointing to the backup path. This module does not intercept packets where Path SelectionVar is 0x00 or 0x01, as the latter bypasses this queue. Intercepted packets are sequentially written to a first-in, first-out local packet buffer queue. This queue is a circular buffer with a preset maximum length of 1024 packets to prevent memory overflow. Each packet is enqueued, and its system timestamp is recorded. .

[0068] After the waiting time in the local packet sending cache queue reaches the dynamic compensation latency value, the packet is then sent to the backup path for transmission.

[0069] A resident send daemon thread continuously checks the head-of-line data packets. Calculate the time it has been waiting in the queue. .like ,in If the dynamic compensation delay value for the current scheduling cycle is used, then the process will remain suspended without any operation. Once... Immediately pop the head packet from the queue and call the underlying send interface to send it to the network card queue corresponding to the backup path for transmission. To ensure timeliness, the daemon thread's check cycle should be less than [time period missing]. The minimum resolution is set to 1 millisecond in this embodiment.

[0070] By introducing a "suspend-wait" sending queue based on dynamic compensation delay value through the above technical solution, the problem of out-of-order data packets caused by the difference in transmission delay of primary and backup paths in multi-path proportional traffic splitting mode is solved, and the order-preserving transmission effect of data streams in multi-path transmission is achieved in non-extreme interference scenarios.

[0071] Furthermore, if the queue compensation mechanism is still used after the "concurrency circuit breaker mode" is determined to be entered, it will introduce intolerable additional latency; if only proportional traffic splitting is performed, the collapse of the main path will result in the loss of more than half of the data. Conventional solutions are completely ineffective in this scenario.

[0072] For example, in the instant a die jamming failure occurs on a stamping line, the equipment current jumps from the rated value to the short-circuit current level within 2 milliseconds, generating a strong electromagnetic pulse. This causes the signal-to-noise ratio of all wireless communication links, including the primary and backup paths, to drop to the noise floor level within 5 milliseconds, resulting in a catastrophic scenario where the success rate of regular data packet transmission is less than 10%.

[0073] To ensure the deterministic transmission of core control commands in catastrophic scenarios where extreme physical shocks and end-to-end network deterioration lead to the complete failure of conventional communication scheduling mechanisms, the specific processing steps of the concurrent circuit breaker mode include:

[0074] Set the dynamic compensation latency value to zero and clear the local packet sending cache queue.

[0075] The system first performs a forced reset operation: resetting the defined dynamic compensation delay value. Clearing the global variable to zero immediately releases all pending data packets. Then, the kernel function `skb_queue_purge` is called to clear the local packet sending buffer queue. This releases the buffer of all queued but unsent packets. This action aims to eliminate any form of ordered waiting, preparing for zero-latency transmission.

[0076] The allocation effect of traffic allocation weights is suspended, and a full clone is performed on the data stream to be sent to obtain a redundant copy sequence.

[0077] The system will calculate the traffic allocation weights from the core infrastructure solution. The logical effect is globally suspended, meaning subsequent packet processing no longer references the proportional splitting instructions in Path Selection Var. Then, the `rte_pktmbuf_clone` function provided by the data plane development kit is called to perform a full clone of each packet in the data stream to be sent. If the original data stream has... A data packet, which, after cloning, generates Each original data packet and the cloned copy constitute a redundant sequence of independent memory copies. .

[0078] Modify the path selection variable in the path scheduling label to a full path broadcast variable, construct a fully redundant scheduling label, and configure it for the redundant replica sequence.

[0079] For each packet in the redundant replica sequence, including both the original and cloned packets, the system forcibly modifies the Path Selection Var variable in its metadata to a newly defined value: 0x03, which is defined as a full-path broadcast variable. Then, a new data structure, "Fully Redundant Scheduling Tag," is constructed, containing only one member: path_mask=0x03, and this tag is assigned to each packet in the redundant replica sequence.

[0080] The multi-path splitter executor synchronously maps redundant replica sequences to the primary and backup communication paths based on the full redundancy scheduling tags to perform zero-latency concurrent transmission.

[0081] The multipath splitting executor reads the full redundancy scheduling label of the data packet and obtains path_mask=0x03. Without performing any weight calculations or queue suspension, it immediately performs deterministic distribution on the redundant replica sequence: the original data packet is sent through the primary path's send queue eth0, while its cloned copy is sent through the backup path's send queue eth1. These two sending operations are completed within the same function call, achieving nanosecond-level synchronization. The receiving end performs duplicate packet detection and elimination: it maintains a sequence number sliding window, delivering the first arriving data packet, regardless of its path, to the upper-layer protocol stack, while discarding subsequent duplicate copies with the same sequence number.

[0082] Simultaneously monitor the cross-domain health coupling index. If the cross-domain health coupling index is lower than the preset safety recovery threshold within the micro-window of the preset count in the current scheduling cycle, it is determined that the physical risk has been eliminated and the concurrent circuit breaker mode is lifted.

[0083] During concurrent circuit breaker mode execution, the system continuously monitors the cross-domain health coupling index defined in the core infrastructure solution. Set a counter. Initially set to 0. At the end of each micro-window, check the average value within that window. Is it below the preset safe recovery threshold? If it is lower, then Add 1; otherwise, Reset to zero. The preset microwindow count threshold is set to zero. .when When all physical risks have been eliminated within five consecutive micro-windows, the system determines that the interference has ended, thus proactively deactivating the concurrent circuit breaker mode and resuming normal branch processing logic.

[0084] The above technical solution, through full cloning and zero-latency concurrent transmission, solves the core security problem of ensuring that critical control commands are not lost and are transmitted uninterruptedly under extreme physical shocks and network degradation scenarios, achieving deterministic level communication assurance and possessing the ability to automatically detect and recover from risks.

[0085] Furthermore, the technical feature of extracting the zero-crossing discrete component of the second derivative of the real-time load current discrete sequence is refined, clarifying the specific signal processing steps for extracting this component from the original current data. The processing procedure for extracting the zero-crossing discrete component of the second derivative by the component acquisition module includes:

[0086] A reference current sequence is obtained by performing a moving average filter on the discrete sequence of real-time load current.

[0087] This embodiment uses the original real-time load current discrete sequence. Perform a moving average filter. Set the filter window length to... For the output sequence The calculation formula is as follows:

[0088] This is equivalent to a low-pass filter with a cutoff frequency of approximately Hz can effectively filter out high-frequency components caused by inverter switching noise, while retaining low-frequency impact characteristics related to changes in mechanical load.

[0089] Discrete second-order difference calculations are performed on the reference current sequence to obtain the second-order derivative sequence.

[0090] This embodiment is... Perform a discrete second-order difference to approximate its second-order time derivative. The calculation formula is as follows:

[0091] ,in, milliseconds. Denominator sec² is used to convert the difference value into a second derivative with physical dimensions.

[0092] Traverse the second derivative sequence, extract the sampling point indices where the positive and negative signs of the values ​​are reversed, and construct the sampling point indices as the zero-crossing discrete components of the second derivative.

[0093] Traversing the sequence of second derivatives For each From 1 to Inspection conditions If satisfied, it means that in and A sign flip occurs between them, and the index of the zero-crossing point is recorded as follows: All indexes that meet the criteria. The set consists of discrete components with zero crossover of the second derivative. If Consider it as a length of The sequence, then at index Set the value to 1 at one position and 0 at the other positions.

[0094] Furthermore, the technical feature of extracting the first-order decay discrete component of the envelope from the real-time signal-to-noise ratio sequence is refined, clarifying the specific signal processing steps for extracting this component from the original signal-to-noise ratio data. The processing procedure for the component acquisition module to extract the first-order decay discrete component of the envelope includes:

[0095] Perform Hilbert transform analysis on the real-time signal-to-noise ratio sequence to obtain the analytical amplitude of the signal and construct the envelope sequence.

[0096] For real-time signal-to-noise ratio sequences Perform the Discrete Hilbert Transform. This transform is implemented in the frequency domain: for Performing a fast Fourier transform yields Multiply the positive frequency component by -j and the negative frequency component by +j, then perform an inverse Fourier transform to obtain its Hilbert transform. Then, analyze the signal. The amplitude is the envelope sequence. :

[0097] The first-order backward difference operation is performed on the envelope sequence, and the data points less than zero in the operation result are extracted as the first-order decaying discrete components of the envelope.

[0098] This embodiment addresses the envelope sequence. Perform a first-order backward difference operation, the formula is:

[0099] , The sign of the envelope indicates the trend of change: a positive value indicates an increase, i.e., an improvement in the signal-to-noise ratio (SNR), while a negative value indicates a decrease, i.e., a deterioration in the SNR. (Torrent traversal) From 1 to Extract all that satisfy Data points. Index these points. and the corresponding decrease value These are recorded and together constitute the first-order decay discrete component of the envelope. If If considered as a sequence, then at the index The value at that location is set to The remaining positions are 0.

[0100] Furthermore, a specific mathematical formula for calculating the cross-domain health coupling index is given, which can be used to accurately quantify the coherence strength between equipment load impact and communication link attenuation.

[0101] The coupling index processing module calculates the cross-domain health coupling index using the following formula:

[0102] ,in, Indicates the cross-domain health coupling index. The second derivative zero-crossing discrete component, For the first-order decay discrete component of the envelope, This is the total number of sampling points within the preset observation window. This is a discrete sampling time index. The range of values ​​for the discrete sampling time index corresponds to the set of sampling points within a preset observation window. The discrete displacement parameter is the iterative variable used to optimize the translation of the second-order derivative zero-crossing discrete component relative to the first-order decaying discrete component of the envelope when calculating the cross-correlation function. To obtain the maximum peak function for cross-correlation calculation, in engineering implementations, to improve computational efficiency, the Fast Fourier Transform is typically used to calculate the cross-correlation. ,Then .

[0103] Furthermore, a specific nonlinear mapping formula for generating the virtual impedance factor is given, which is used to transform the cross-domain health coupling index into a scheduling cost factor that grows superlinearly with the degree of risk.

[0104] The factor processing module generates the virtual impedance factor using the following formula:

[0105] ,in, Represents the virtual impedance factor. This represents the real-time packet loss rate of the main communication path at the current discrete sampling time index. As a cross-domain health coupling index, It is a natural exponential function. This is the preset sensitivity correction factor. The formula uses an exponential function. Achieved packet loss rate The nonlinear gain.

[0106] For example: Let ,when At that time, the gain is 1. ;when At that time, gain , Expanded by 2.7 times; when At that time, gain , This represents a 7.4-fold increase. This superlinear growth ensures that the system can generate a risk signal far exceeding the conventional packet loss rate indication the instant physical disturbances and communication degradation begin to couple.

[0107] Furthermore, a specific Sigmoid normalized mapping formula from the positive offset to the flow allocation weight is given to achieve smooth, non-linear flow ratio adjustment.

[0108] The communication scheduling module calculates the traffic allocation weight using the following formula:

[0109] ,in, Assign weights to traffic. This is a positive offset. The preset weight transition smoothing coefficient is used, and , It is a natural exponential function. It also represents the proportion of data packets allocated to the backup path. This function will use the positive offset. Mapped to The interval. Its derivative is in The nearest maximum ensures the system's sensitive response when the risk just exceeds the threshold. When When the value increases from 0 to 2, the Sigmoid function value rises rapidly from 0.5 to approximately 0.88; while when As the function value continues to increase, it approaches 1, achieving soft saturation of the traffic splitting ratio. This avoids instantly switching all traffic to a backup path that may also be disrupted under extreme risk conditions, thus maintaining the robustness of the system.

[0110] This application also provides a computer-readable storage medium for storing a computer program, which, when executed by a processor, implements a smart manufacturing production line data communication scheduling system based on edge computing.

[0111] The storage medium can be any type of non-volatile memory, such as read-only memory, programmable read-only memory, erasable programmable read-only memory, electrically erasable programmable read-only memory, flash memory, solid-state drive, hard disk drive, etc.

[0112] For example, when executed on an x86-based server or an ARM-based embedded edge computing node deployed at the edge of a smart manufacturing production line, it can implement all the functional modules and steps of any one or a combination of the above-mentioned solutions for an edge computing-based smart manufacturing production line data communication scheduling system. By deploying this storage medium on an edge computing node or production line control server, the scheduling system of this invention can be easily deployed and upgraded in software without modifying the hardware circuit, thus reducing engineering implementation costs and system maintenance complexity.

[0113] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0114] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0115] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0116] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0117] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0118] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A smart manufacturing production line data communication and scheduling system based on edge computing, characterized in that, include: Component acquisition module: used to simultaneously acquire the real-time load current discrete sequence of the target device node and the real-time signal-to-noise ratio sequence of the main communication path within a preset observation window, extract the second derivative zero-crossing discrete component of the real-time load current discrete sequence, and the envelope first-order attenuation discrete component of the real-time signal-to-noise ratio sequence. Coupling index processing module: Used to perform coherence analysis on the second derivative zero-crossing discrete components and the first-order decaying discrete components of the envelope with the same sampling frequency using the cross-correlation algorithm, and to obtain the cross-domain health coupling index. Factor processing module: Used to generate a virtual impedance factor by weighting the real-time packet loss rate of the main communication path with the cross-domain health coupling index as the independent variable and through a preset exponential gain function; Offset processing module: used to determine whether the virtual impedance factor is greater than the preset impedance threshold. If it is greater, it sends a context pre-synchronization command to the backup gateway and calculates the positive offset of the virtual impedance factor exceeding the preset impedance threshold. Communication scheduling module: It is used to input the positive offset into the Sigmoid normalization function, calculate the traffic allocation weight, and configure the path scheduling label containing path selection variables for the data stream to be sent. Through the multi-path flow splitter, it performs data stream path mapping transmission between the main communication path and the backup path based on the path selection variables in the path scheduling label, and outputs the standardized data scheduling result of the current scheduling period.

2. The intelligent manufacturing production line data communication and scheduling system based on edge computing according to claim 1, characterized in that, Before performing data flow path mapping and transmission between the primary and backup communication paths based on traffic allocation weights, concurrent conflict determination and scheduling reconfiguration are also performed: The preset observation window is divided into several micro windows, and a synchronous mapping relationship between the scheduling cycle of the virtual impedance factor and the micro windows is established, wherein each scheduling cycle contains at least one micro window. Obtain the synchronous mapping relationship between the scheduling period of the virtual impedance factor and the micro-window of the real-time load current discrete sequence. Calculate the average value of the moving variance of the discrete sequence of real-time load current for all micro windows within the current scheduling period, and normalize it to obtain the normalized average moving variance, which is then recorded as the transient impact characteristic. Calculate the discrete absolute value of the virtual impedance factor in the current scheduling period relative to the previous scheduling period, and denote it as the degradation gradient feature characterizing the rate of communication degradation. The product of the degradation gradient feature and the preset time delay conversion coefficient is obtained to get the dynamic compensation time delay value; The transient impact characteristics and degradation gradient characteristics are weighted and summed to obtain the comprehensive conflict evaluation index, and it is determined whether the comprehensive conflict evaluation index is greater than the preset conflict threshold. If it is not greater than, then the multi-path shunt executor remains in out-of-order compensation mode; If the value is greater than the specified value, the multipath cascading executor will be forced to switch to concurrent circuit breaker mode.

3. The intelligent manufacturing production line data communication and scheduling system based on edge computing according to claim 2, characterized in that, The specific processing steps of the disordered compensation mode include: Intercept packets containing path scheduling tags pointing to backup paths and suspend them by writing them to the local packet sending cache queue; After the waiting time in the local packet sending cache queue reaches the dynamic compensation latency value, the packet is then sent to the backup path for transmission.

4. The intelligent manufacturing production line data communication and scheduling system based on edge computing according to claim 2, characterized in that, The specific processing steps of the concurrent circuit breaker mode include: Set the dynamic compensation latency value to zero and clear the local packet sending cache queue; The allocation effect of the traffic allocation weight is suspended, and a full clone is performed on the data stream to be sent to obtain a redundant copy sequence; Modify the path selection variable in the path scheduling label to a full path broadcast variable, construct a fully redundant scheduling label, and configure it for the redundant replica sequence; The multi-path splitter executor synchronously maps redundant replica sequences to the main communication path and backup path based on the full redundancy scheduling tag to perform zero-latency concurrent transmission; Simultaneously monitor the cross-domain health coupling index. If the cross-domain health coupling index is lower than the preset safety recovery threshold within the micro-window of the preset count in the current scheduling cycle, it is determined that the physical risk has been eliminated and the concurrent circuit breaker mode is lifted.

5. The intelligent manufacturing production line data communication and scheduling system based on edge computing according to claim 1, characterized in that, The process of extracting the second-order derivative zero-crossing discrete components by the component acquisition module includes: A reference current sequence is obtained by performing a moving average filter on the real-time discrete load current sequence; Discrete second-order difference calculations are performed on the reference current sequence to obtain the second-order derivative sequence; Traverse the second derivative sequence, extract the sampling point indices where the positive and negative signs of the values ​​are reversed, and construct the sampling point indices as the zero-crossing discrete components of the second derivative.

6. The intelligent manufacturing production line data communication and scheduling system based on edge computing according to claim 1, characterized in that, The process of the component acquisition module extracting the first-order attenuation discrete component of the envelope includes: Perform Hilbert transform analysis on the real-time signal-to-noise ratio sequence to obtain the analytical amplitude of the signal and construct the envelope sequence; Perform a first-order backward difference operation on the envelope sequence, and extract the data points less than zero in the result as first-order decaying discrete components of the envelope.

7. The edge computing-based intelligent manufacturing production line data communication scheduling system according to claim 5 or 6, characterized in that, The coupling index processing module calculates the cross-domain health coupling index using the following formula: ,in, Indicates the cross-domain health coupling index. The second derivative is a zero-crossing discrete component. For the first-order decay discrete component of the envelope, The first-order decay discrete component of the envelope after discrete displacement translation. This is the total number of sampling points within the preset observation window. This is a discrete sampling time index. The range of values ​​for the discrete sampling time index corresponds to the set of sampling points within a preset observation window. The discrete displacement parameter is the iterative variable used to optimize the translation of the second-order derivative zero-crossing discrete component relative to the first-order decaying discrete component of the envelope when calculating the cross-correlation function. This is the function that maximizes the peak value of the cross-correlation operation.

8. The intelligent manufacturing production line data communication and scheduling system based on edge computing according to claim 7, characterized in that, The factor processing module generates the virtual impedance factor using the following formula: ,in, Represents the virtual impedance factor. This represents the real-time packet loss rate of the main communication path at the current discrete sampling time index. As a cross-domain health coupling index, It is a natural exponential function. This is the preset sensitivity correction factor.

9. The intelligent manufacturing production line data communication and scheduling system based on edge computing according to claim 1, characterized in that, The communication scheduling module calculates the traffic allocation weight using the following formula: ,in, Assign weights to traffic. This is a positive offset. The preset weight transition smoothing coefficient is used, and , It is a natural exponential function.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the system as described in any one of claims 1-9.

Images