An aluminum alloy ingredient optimization and composition adjustment system

Abstract

This application relates to the technical field of alloy batching control, and in particular to an aluminum alloy batching optimization and composition adjustment system, which includes a field sensing terminal, an edge control gateway, and an execution terminal. The field sensing terminal reports weight messages to the edge control gateway; the edge control gateway extracts the feeding flow rate and obtains channel quality indication and retransmission count values; the edge control gateway calculates the estimated dwell time delay and calculates the overshoot compensation amount to reset the preset target threshold to a dynamic pre-truncation threshold; the edge control gateway has a built-in asynchronous threshold comparator that calculates the current transient weight; when the weight contained in the weight message or the transient weight reaches the dynamic pre-truncation threshold, the edge control gateway sends a pre-truncation signaling command to the execution terminal; the execution terminal drives the batching valve to close according to the pre-truncation signaling command. This application can overcome batching overshoot caused by network latency and dynamically reset the truncation threshold based on the air interface status, thereby achieving cross-layer physical compensation.

Description

Technical Field

[0001] This application relates to the technical field of alloy batching control, and in particular to an aluminum alloy batching optimization and composition adjustment system. Background Technology

[0002] Aluminum alloy batching and composition adjustment are core processes in metal smelting, and their accuracy directly determines the physical properties and economic benefits of the final alloy material. As modern industrial process control architectures evolve towards distributed systems, the dynamic and continuous physical processes of high-speed material feeding on-site are increasingly relying on digital data communication networks and edge computing technologies for data aggregation and centralized management. Under this edge computing control architecture, the underlying sensing devices at the batching site must continuously and frequently upload extremely time-sensitive weight status messages. These messages are then centrally processed by remote control nodes, which then issue execution commands to drive the timing of actions of the underlying servo mechanisms and achieve precise cutting of the fluid material.

[0003] Existing technologies have explored calculation and control methods for alloy batching processes. For example, Chinese invention patent CN113742924A discloses a batching method for smelting special alloy aluminum ingots. This technical solution mainly extracts the set parameters of the batching process, constructs a multi-objective optimization model that takes into account factors such as the total cost of raw material input, and runs a multi-objective non-dominated sorting optimization algorithm incorporating distributed estimation during the solution stage. Iterative calculations are performed through normal distribution sampling of multiple generations of populations and reference point association screening to output an optimized batching scheme for special alloy aluminum ingots that conforms to static production experience and total quantity constraints.

[0004] Static algorithm models built on ideal execution assumptions reveal deep-seated limitations in real-time dynamic control scenarios relying on edge computing and wireless network communication. In industrial settings with strong electromagnetic interference and radio frequency noise from the start-up and shutdown of large equipment, the quality of wireless communication channels carrying industrial control flows is highly susceptible to random fading. Limited by the inherent architecture of existing technologies, which are purely based on static batching planning and ideal timing assumptions, their control logic lacks a perception dimension of the non-deterministic latency jitter of the underlying communication link. When high-speed fluid feeding in the physical space and data command transmission in the network space operate in parallel, if interference in the communication channel causes a large number of queues or data retransmissions of the underlying control frames, the valve shut-off control signal issued by the application layer will inevitably experience unpredictable dwell time delays within the network transmission layer. The statically fixed batching algorithm and triggering mechanism cannot predict and adapt to this sudden network communication time distortion, inevitably causing the actual shut-off action of the underlying mechanical valve to lag behind the theoretical trigger time. Liquid aluminum alloy in a high-speed jet state continues to be injected into the molten pool within this delay difference, ultimately leading to severe physical weight overshoot and batching exceeding limits. Summary of the Invention

[0005] To overcome the overshooting of batching caused by network latency, this application provides an aluminum alloy batching optimization and composition adjustment system that dynamically resets the cutoff threshold based on the air interface status to achieve cross-layer physical compensation.

[0006] This application provides an aluminum alloy batching optimization and composition adjustment system, which adopts the following technical solution: An aluminum alloy batching optimization and composition adjustment system includes a field sensing terminal, an edge control gateway, and an execution terminal; The field sensing terminal reports weighted messages with timestamps to the edge control gateway at a preset period. The edge control gateway extracts the unloading flow rate based on the continuously received weight messages, and obtains the channel quality indicator and retransmission count value of the wireless link where the execution terminal is located. The edge control gateway calculates the estimated dwell time based on the channel quality indicator and the retransmission count value, and calculates the overshoot compensation amount based on the estimated dwell time and the material flow rate, so as to reset the preset target threshold to the dynamic pre-truncation threshold. The edge control gateway has a built-in asynchronous threshold comparator, which calculates the current transient weight by combining the received weight message with the material flow rate. When the weight contained in the weight message or the transient weight reaches the dynamic pre-truncation threshold, the edge management gateway sends a pre-truncation signaling carrying a priority flag to the execution terminal. The execution terminal triggers a hardware interrupt based on the pre-truncation signaling to drive the batching valve to close.

[0007] Optionally, the edge control gateway includes a flow rate observation module and a local register; The flow velocity observation module extracts a sequence of messages that meet a preset duration threshold from the continuously received weight messages based on a sliding time window. The flow rate observation module performs smoothing filtering on the message sequence to extract the weight trend component, and calculates the feeding flow rate based on the weight trend component; The flow rate observation module writes the feeding flow rate into the local register, which is then read by the edge control gateway to calculate the overshoot compensation amount.

[0008] Optionally, the field sensing terminal interacts with the edge control gateway to perform master-slave clock synchronization in order to extract the clock offset; The edge control gateway uses the clock offset to compensate for the timestamp carried in the weight message to generate a calibration timestamp; The flow velocity observation module defines the true span of the sliding time window based on the extracted calibration timestamp, and uses this as a comparison benchmark to determine whether the message sequence meets the preset duration threshold.

[0009] Optionally, the edge management gateway has a pre-set mapping table for basic air interface frame length, single retransmission time, and backoff compensation. The edge management gateway obtains the channel quality indicator and the retransmission count value through the wireless network information service interface; The edge management gateway retrieves the queue backoff compensation amount from the backoff compensation mapping table using the channel quality indicator as an index, and sums the product of the basic air interface frame length, the retransmission count value and the single retransmission time, and the queue backoff compensation amount to calculate the estimated dwell time.

[0010] Optionally, when the retransmission count value is greater than a preset reference threshold or the drop value of the channel quality indicator in an adjacent period reaches a preset attenuation order, the edge management gateway triggers the calculation of the estimated dwell time. The edge control gateway multiplies the estimated dwell time by the material flow rate to obtain the overshoot compensation amount, and subtracts the overshoot compensation amount from the preset target threshold to reset the dynamic pre-truncation threshold.

[0011] Optionally, the asynchronous threshold comparator operates independently of the reception cycle of the weight message; The time difference between the current system time and the timestamp carried in the most recently received weight message is calculated. The transient weight is obtained by linear extrapolating the product of the time difference and the feed flow rate, plus the weight contained in the weight message.

[0012] Optionally, the execution terminal includes a communication module, an internal bus, and a programmable logic controller; The edge management gateway configures the network layer header of the pre-truncation signaling with a preset priority to form the priority tag; The communication module parses the pre-truncation signaling carrying the priority flag and triggers the hardware interrupt to the programmable logic controller via the internal bus; The programmable logic controller responds to the hardware interrupt to shut off the dispensing valve.

[0013] Optionally, the edge control gateway has a response masking timer and a constant flow rate; When the channel quality indicator is less than the preset deep fading threshold, the edge control gateway sends a physical throttling signal to drive the execution terminal to adjust the batching valve to the preset opening degree and start the response shielding timer to freeze the reset of the dynamic pre-truncation threshold; After the response shielding timer reaches a preset duration, the edge control gateway recalculates the overshoot compensation amount by replacing the feed flow rate with the constant flow rate, thereby resetting the dynamic pre-truncation threshold.

[0014] Optionally, after the edge control gateway issues the physical throttling signaling, if the number of times the channel quality indication obtained continuously exceeds the preset recovery threshold reaches a preset stable period, then the edge control gateway issues a throttling release signaling to the execution terminal and recalculates the overshoot compensation amount by replacing the constant flow rate with the material feeding flow rate.

[0015] Optionally, when the edge control gateway does not receive the weight message for a duration that reaches a preset disconnection span threshold, the difference between the current system time and the calibration timestamp corresponding to the most recently received weight message is calculated as a time domain increment. The product of the time domain increment and the unloading flow rate extracted from the local register is added to the weight contained in the weight message to obtain a virtual inferred weight. Then, when the virtual inferred weight reaches the dynamic pre-truncation threshold, the pre-truncation signaling is sent to the execution terminal.

[0016] In summary, this application includes the following beneficial technical effects: 1. The edge control gateway calculates the estimated dwell time based on the wireless link channel quality indicator and retransmission count value, and determines the overshoot compensation amount based on the product of the estimated dwell time and the material flow rate to reset the dynamic pre-truncation threshold. This allows the valve shut-off triggering time to adaptively advance with the fluctuation of network transmission delay, compensating for the action lag caused by the nondeterministic delay of the communication layer from the physical control level, and avoiding the overshoot phenomenon of material feeding caused by the deterioration of the air interface state.

[0017] 2. By performing master-slave clock synchronization between the field sensing terminal and the edge control gateway to obtain the clock offset and compensate for the timestamp of the weight message, combined with the mechanism of the flow rate observation module to extract the weight trend component based on the sliding time window and smoothing filter, the real-time calculation accuracy of the feeding flow rate is improved under the global unified time reference. This provides a more accurate physical quantity reference for resetting the dynamic pre-truncation threshold and reduces the calculation error of the compensation amount caused by time deviation and weight noise.

[0018] 3. By using an asynchronous threshold comparator to operate independently of the weight message receiving cycle, and by using the difference between the current system time and the calibration timestamp of the most recently received message, as well as the material flow rate to linearly extrapolate the transient weight, continuous virtual sensing and threshold cross-detection of the molten pool weight are achieved within the message reporting interval. This avoids the blind spot in the truncation command response caused by the discreteness of message reporting and ensures the timeliness of the threshold trigger response. Attached Figure Description

[0019] Figure 1This is a schematic diagram of the overall system structure according to an embodiment of this application; Figure 2 This is a graph showing the cross-layer air interface status perception and dynamic pre-truncation threshold reset in an embodiment of this application. Detailed Implementation

[0020] The following combination Figures 1-2 This application will be described in further detail.

[0021] This application discloses an aluminum alloy batching optimization and composition adjustment system, applied to the quantitative batching and precise composition adjustment of liquid materials in the aluminum alloy smelting process. The system is implemented based on a 5G industrial wireless edge network architecture and comprises three core functional nodes: a field sensing terminal, an edge control gateway, and an execution terminal. The supporting network equipment is a 5G base station, and all nodes possess microsecond-level time synchronization capabilities.

[0022] The on-site sensing terminal is deployed below the molten aluminum pool, responsible for collecting and reporting real-time weight data of the liquid aluminum alloy in the pool. The edge control gateway is deployed on a multi-access edge computing platform (MEC platform), serving as the core computing and control unit of the system. It integrates a flow rate observation module, local registers, a wireless network information service parsing engine, and an asynchronous threshold comparator. The 5G base station is responsible for establishing and maintaining the air interface, providing the edge control gateway with interfaces for querying and subscribing to the underlying air interface status. The execution terminal is deployed at the end of the batching pipeline, integrating a batching valve, a valve position feedback unit, an emergency shut-off solenoid valve, a local storage unit, and a timing unit. The emergency shut-off solenoid valve is connected in series with the main batching valve along the discharge pipeline, located upstream of the main batching valve. The core structure of the execution terminal includes a communication module, an internal bus, and a programmable logic controller (PLC). The valve position feedback unit, emergency shut-off solenoid valve, local storage unit, and timing unit are all connected to the PLC via the internal bus. The communication module interacts with the PLC via the internal bus. The execution terminal as a whole is responsible for receiving control commands and performing precise valve shut-off control.

[0023] like Figure 1As shown, an aluminum alloy batching optimization and composition adjustment system establishes a wireless communication link based on a 5G base station. Its physical topology includes a field sensing terminal, an edge control gateway, and an execution terminal. In terms of static connection configuration, the field sensing terminal is positioned at the bottom side of the batching molten pool, the edge control gateway is centrally deployed within a multi-access edge computing (MEC) platform, and the execution terminal is located at the end of the liquid aluminum alloy batching pipeline. The main batching valve and emergency shut-off solenoid valve under the execution terminal are installed in series along the axial direction of the feed pipeline, with the emergency shut-off solenoid valve located upstream of the main batching valve. In terms of dynamic interactive working principle, the field sensing terminal and the edge control gateway act as slave and master clock nodes, respectively. Through the 5G base station air interface resources, they periodically exchange synchronization messages and delay request messages according to the IEEE 1588v2 protocol to establish a globally unified time reference. Under this reference, the field sensing terminal continuously sends weighted messages carrying timestamps to the uplink. The edge control gateway, after calculation, distributes pre-truncation instructions to the downlink to drive the execution terminal hardware action. The following steps will be described in detail: S1 System Network Deployment and Global Master-Slave Clock Synchronization The initial input conditions for this step are that the hardware deployment and physical networking of each core functional node of the system are completed, and basic wireless communication links are established between the field sensing terminal, edge management gateway, 5G base station and execution terminal, and basic data interaction capabilities are available.

[0024] After the system network is completed, the field sensing terminals and the edge management gateway use the IEEE 1588v2 precise time protocol for master-slave clock synchronization. The synchronization period is set to 1 millisecond, consistent with the reporting period of weighted messages. The edge management gateway acts as the master clock node, and the field sensing terminals act as slave clock nodes. The master clock node sends a synchronization message to the slave clock node, carrying the master clock node's message transmission time. The slave clock node records the local reception time of the synchronization message and then sends a delay request message to the master clock node, carrying the slave clock node's message transmission time. The master clock node records the local reception time of the delay request message and sends this reception time back to the slave clock node. Based on the four time values—synchronization message transmission time, synchronization message reception time, delay request message transmission time, and delay request message reception time—the slave clock node calculates the average delay of the bidirectional transmission path and the clock offset between the master and slave nodes, controlling the synchronization accuracy to within 1 microsecond. Through the full-process delay calculation and offset calibration of the above two-way message interaction, the clock reference of the master and slave nodes is accurately aligned, eliminating the time reference error caused by transmission path asymmetry and node crystal oscillator deviation.

[0025] After obtaining the clock offset returned from the clock node, the edge control gateway performs an absolute value check on the clock offset to ensure that the clock synchronization accuracy error between the master and slave nodes does not exceed 1 microsecond. It then uses this clock offset to compensate for the original timestamp carried in subsequently received weight messages, generating a calibration timestamp. The calibration timestamp is perfectly aligned with the edge control gateway's local system time base, providing a unified time scale for end-to-end data processing. It has a direct derived correspondence with the original timestamp carried in the weight message, and the two have a unique mapping relationship.

[0026] After the on-site sensing terminal completes master-slave clock synchronization, it continuously reports weight messages to the edge control gateway at a preset period. The preset period is set to 1 millisecond. This period is based on the fact that the maximum feeding rate of aluminum alloy liquid material is 50 kg / s, and the weight change within a 1-millisecond period does not exceed 0.05 kg, which can meet the accuracy requirement of ±0.1 kg for industrial batching, and is also compatible with the micro-slot scheduling period of 5G ultra-reliable low-latency communication services.

[0027] The payload of the weight message includes the message sequence number, the original timestamp, and the real-time weight value. The edge control gateway continuously monitors the master-slave clock synchronization status. If the clock offset exceeds the synchronization accuracy threshold of 1 microsecond for three consecutive synchronization cycles, clock synchronization is deemed to have failed. A synchronization failure alarm is immediately sent to the field sensing terminal and the execution terminal, freezing the material flow rate calculation and resetting the dynamic pre-truncation threshold. Simultaneously, the execution terminal is instructed to close the batching valve to a safe opening of 10%. After this step is completed, the system establishes a globally unified time base, possessing a stable and continuous weight data reporting capability with a precise time scale, providing complete basic data support for subsequent calculations.

[0028] Sliding window observation and stable extraction of S2 feed flow rate The initial input condition for this step is that the edge control gateway has completed master-slave clock synchronization with the field sensing terminal, established a globally unified time reference, and continuously acquired weight messages with calibration timestamps reported by the field sensing terminal.

[0029] This step is performed by the flow rate observation module built into the edge control gateway. The flow rate observation module uses a sliding time window to extract a sequence of messages from continuously received weight messages that meets a preset duration threshold. The preset duration threshold is set to 10 milliseconds. This threshold is set because a 10-millisecond duration can cover at least 10 consecutively reported weight messages, effectively smoothing out high-frequency weight noise caused by molten pool surface sloshing, while fully preserving the true trend of the feed flow rate. This avoids drastic fluctuations in flow rate calculation results due to an excessively short window duration, or delayed response to flow rate changes due to an excessively long window duration. The sliding step size of the sliding time window is 1 millisecond, consistent with the reporting cycle of the weight messages. Each time a new weight message is received, the window slides forward one message cycle, completing one update of the message sequence. The flow rate observation module defines the true span of the sliding time window based on the calibration timestamp, using this as the comparison benchmark to determine whether the message sequence meets the preset duration threshold, eliminating the influence of master-slave node clock deviation on the window duration definition.

[0030] The flow rate observation module first performs a 3-point first-order moving average smoothing filter on the real-time weight values ​​of the message sequence within the sliding time window to extract the weight trend component. Then, it sorts the filtered weight trend component according to the order of the calibration timestamps and calculates the ratio of the weight difference between two adjacent sets of data to the calibration timestamp difference to obtain the single-step flow rate value. The physical meaning of the single-step flow rate value is the instantaneous feeding speed within the interval between two adjacent weight acquisitions, with the dimension of kilograms per second, which is completely consistent with the physical dimension of the final output feeding flow rate. The flow rate observation module performs an arithmetic average of all single-step flow rate values ​​within the window to obtain the final feeding flow rate.

[0031] The flow rate monitoring module verifies the validity of the calculated material discharge rate. If the discharge rate is negative or exceeds the maximum design flow rate of 50 kg / s, the flow rate value is deemed abnormal, the calculation result is discarded, and the valid discharge rate of the previous cycle latched in the local register is used. Simultaneously, an abnormal flow rate event is reported to the edge control gateway. After successful verification, the flow rate monitoring module writes the calculated discharge rate into the local register of the edge control gateway. The local register uses a latching mechanism, updating the value only after completing the full calculation of the complete sliding time window, and does not write intermediate calculation results during the window sliding process.

[0032] After this step is completed, the system obtains a stable and effective material flow rate under the current working conditions, providing a core physical benchmark for subsequent calculations.

[0033] S3 Cross-Layer Air Interface Status Awareness and Dynamic Pre-Truncation Threshold Reset The initial input conditions for this step are that the edge control gateway has completed the sliding window observation and stable extraction of the material flow rate, the local register stores the stable and effective material flow rate under the current working conditions, the system maintains a globally unified time base, continuously acquires the weight messages reported by the field sensing terminals, and maintains a stable communication connection with the 5G base station.

[0034] This step is the core computational component of the system, executed by the built-in Wireless Network Information Service (WLANIS) parsing engine of the edge management gateway. The edge management gateway continuously subscribes to the air interface status parameters of the downlink wireless link where the executing terminal is located through the WLANIS interface opened by the 5G base station, obtaining the Channel Quality Indicator (CQI) and the hybrid automatic repeat request (ARP) retransmission count value from the Media Access Control (MAC) layer. The obtained air interface status parameters are cached in real time, with the caching period consistent with the reporting period of weight packets, ensuring complete alignment between the air interface status parameters and the material unloading rate in the time dimension.

[0035] The edge management gateway has a pre-built mapping table for basic air interface frame length, single retransmission time, and backoff compensation. The basic air interface frame length is set to 0.125 milliseconds, a value adapted to the micro-slot configuration of 5G ultra-reliable low-latency communication services, representing the minimum transmission time for a single control signal. The single retransmission time is set to 1 millisecond, a value representing the typical round-trip latency of hybrid automatic repeat request retransmission as defined by the 3GPP protocol, covering the entire process time of retransmission scheduling and data transmission. The backoff compensation mapping table is based on the correspondence between CQI and modulation and coding strategies defined in the 3GPP TS 38.214 protocol. It uses CQI as the core index and can also set auxiliary indexes based on other air interface status parameters such as retransmission count and channel occupancy rate. The table clarifies the queue backoff compensation amount corresponding to different index parameters. The higher the CQI value, the smaller the corresponding queue backoff compensation amount; the two are negatively correlated. Typical mapping examples are: CQI = 1, queue backoff compensation amount is 5 milliseconds; CQI = 3, queue backoff compensation amount is 2 milliseconds; CQI = 7, queue backoff compensation amount is 0.5 milliseconds; CQI = 15, queue backoff compensation amount is 0.1 milliseconds.

[0036] The edge control gateway uses the acquired CQI as an index to retrieve the corresponding queue backoff compensation amount from the backoff compensation mapping table. It then sums the product of the basic air interface frame length, the retransmission count, and the single retransmission time, along with the queue backoff compensation amount, to calculate the estimated dwell time. The estimated dwell time fully covers the entire transmission time from the edge control gateway's issuance to the execution terminal's reception, including the basic transmission time, additional retransmission time, and queuing backoff time caused by channel congestion. All calculated items are measured in milliseconds, ensuring consistent dimensions in the calculation results.

[0037] The edge control gateway compares channel state parameters every 1 millisecond. When the retransmission count exceeds a preset baseline threshold, or the CQI drop in an adjacent period reaches a preset attenuation level, it triggers a recalculation of the estimated dwell time and a reset of the dynamic pre-truncation threshold. The preset baseline threshold is set to 2. This threshold is based on the fact that a single retransmission corresponds to an additional 1 millisecond of time consumption. The latency increase caused by two consecutive retransmissions exceeds the acceptable valve action lag threshold of the batching system, which will cause observable weight overshoot. The preset attenuation level is set to 3. This level is based on the fact that a single CQI drop of more than 3 levels indicates a sudden deep fading of the channel state, and the transmission latency will experience an unpredictable surge, requiring an immediate reassessment of the latency level. Even if the above triggering conditions are not met, the edge control gateway periodically updates the estimated dwell time every 10 milliseconds, simultaneously resetting the dynamic pre-truncation threshold. The periodic update uses the hybrid automatic repeat request retransmission count value statistically collected by the media access control layer within the current period, rather than the cumulative value.

[0038] The edge control gateway reads the current effective feeding rate from its local register, converts the estimated dwell time delay to seconds, and multiplies it by the feeding rate to calculate the overshoot compensation. The physical meaning of the overshoot compensation is the weight increment caused by the continuous feeding of liquid aluminum alloy during control signal transmission; that is, the theoretical overshoot weight caused by valve closing lag. The physical dimension of the calculated result is kilograms, which is completely consistent with the dimension of the weight threshold. The edge control gateway subtracts the overshoot compensation from the preset target threshold, completing the reset of the dynamic pre-truncation threshold. The preset target threshold is the target weight value set by the aluminum alloy batching process. The dynamic pre-truncation threshold is always less than or equal to the preset target threshold and is adjusted synchronously as the channel condition deteriorates, reserving sufficient lead time to offset the overshoot effect caused by transmission delay.

[0039] Reference Figure 2 This demonstrates the principle of cross-layer physical compensation and dynamic threshold adjustment under nondeterministic network delay conditions. In the coordinate system, the horizontal axis represents the progression of the time domain, and the vertical axis represents the cumulative value of the transient weight of the liquid aluminum alloy. Figure 2 The preset target threshold is a constant horizontal benchmark, limiting the static target weight of the batching process. In terms of dynamic adjustment, when the edge control gateway receives a sudden drop in the Channel Quality Indicator (CQI) obtained through the wireless network information service interface, or when the retransmission count exceeds the preset benchmark threshold, the gateway's internal calculation logic triggers a conversion of the estimated dwell time. For example... Figure 2As shown by the step-like downward offset of the curve, the edge control gateway uses the product of the estimated dwell time and the real-time feeding flow rate as the overshoot compensation amount, and performs linear deduction from the preset target threshold, thereby resetting the baseline for issuing the command that triggers the valve closing action to the dynamic pre-truncation threshold. The physical dwell error caused by the queuing and retransmission of control signals in the air interface is offset by the pre-emptive response in the time domain.

[0040] After this step is completed, the system obtains a dynamic pre-truncation threshold that matches the current channel state in real time, providing a triggering benchmark for subsequent threshold comparison and control operations.

[0041] S4 asynchronous threshold cross-checking and high-priority truncation closed-loop execution The initial input conditions for this step are that the edge control gateway has completed cross-layer air interface status perception and dynamic pre-truncation threshold reset, has a dynamic pre-truncation threshold that matches the current channel status in real time, the system maintains a globally unified time base, continuously acquires weight messages with calibration timestamps reported by the field sensing terminal, has a stable and effective material feeding rate under the current working conditions latched in the local register, and the edge control gateway and the execution terminal maintain a stable downlink wireless communication link.

[0042] The asynchronous threshold comparator built into the edge control gateway operates independently of the weight message reception cycle. It uses the edge control gateway's local system clock as its operating reference and is unaffected by weight message reporting frequency, message reception interruptions, or uplink status fluctuations, performing uninterrupted continuous calculations. The asynchronous threshold comparator continuously calculates the time difference between the current system time and the timestamp carried in the most recently received weight message. When calculating the time difference, this "timestamp" must specifically refer to a "calibrated timestamp" after master-slave clock offset compensation and alignment with the gateway's system time. The time difference is then multiplied by the unloading flow rate stored in the local register to obtain the unloading weight increment within the corresponding time interval. This weight increment is then added to the measured weight value contained in the weight message, and the transient weight at the current moment is obtained through linear extrapolation. The asynchronous threshold comparator performs a limiting process with a lower limit of 0 on the calculated time difference value. If the time difference is negative, it is directly taken as 0 as the calculated value, and the transient weight is equal to the most recently received measured weight.

[0043] The asynchronous threshold comparator continuously performs dual-dimensional threshold cross-comparison, synchronously comparing the measured weight, the transient weight obtained by linear extrapolation, and the dynamic pre-truncation threshold values ​​contained in the weight message. When either the measured weight or the transient weight reaches the dynamic pre-truncation threshold, the edge control gateway immediately generates a pre-truncation signaling and sends it to the execution terminal. The edge control gateway has built-in signaling issuance interlocking logic. After the first pre-truncation signaling is issued, an interlocking flag is immediately triggered, and a 5-millisecond signaling response timeout timer is started. If no signaling reception confirmation frame is received from the execution terminal before the timer expires, the interlocking flag is released, and the pre-truncation signaling is reissued; if an confirmation frame is received, the interlocking flag is maintained until the current batching process ends.

[0044] After the pre-truncation signaling is issued, the asynchronous threshold comparator immediately terminates the linear extrapolation calculation of transient weight and switches to threshold comparison based on the measured weight of the packet. The edge management gateway performs priority marking configuration on the generated pre-truncation signaling, configuring the Differentiated Services Code Point field of the network layer header of the pre-truncation signaling to the accelerated forwarding level, corresponding to the binary value 101110. This serves as the preset priority marker, ensuring that the pre-truncation signaling obtains the highest priority scheduling authority in all forwarding nodes of the wireless network, skipping the queuing process of ordinary service data, prioritizing the use of air interface transmission resources, minimizing queuing latency and transmission jitter during signaling transmission, and compressing the end-to-end dwell time of the signaling.

[0045] After receiving the pre-truncation signaling, the communication module of the execution terminal directly identifies the priority flag in the signaling header. For pre-truncation signaling carrying the accelerated forwarding level flag, it performs pass-through forwarding processing, bypassing unnecessary protocol parsing at the application layer and directly triggering a hardware interrupt to the programmable logic controller (PLC) via the internal bus. This hardware interrupt is the highest priority non-maskable interrupt of the PLC, capable of interrupting all currently executing non-security user programs, keeping the interrupt response delay within 1 microsecond. Upon responding to the hardware interrupt, the PLC immediately outputs a constant-amplitude shutdown control signal, directly driving the servo actuator of the batching valve to perform a full-close action, cutting off the liquid aluminum alloy feeding path. After outputting the shutdown control signal, the PLC continuously collects the position feedback signal of the batching valve. If the valve is not fully closed within 20 milliseconds, it immediately triggers an audible and visual alarm, reports valve fault information to the edge control gateway, and simultaneously drives the emergency shut-off solenoid valve to forcibly shut off the feeding path.

[0046] After this step is completed, the system will have completed the precise batching and composition adjustment of aluminum alloy under normal operating conditions.

[0047] S5 Adaptive Control and Self-Healing Backup for Extreme Abnormal Conditions The initial input conditions for this step are that the system has completed asynchronous threshold cross-detection and high-priority truncation closed-loop execution under normal operating conditions, the edge management gateway continuously acquires the CQI and retransmission count value of the downlink wireless link where the execution terminal is located, the field sensing terminal reports weight messages with calibration timestamps to the edge management gateway according to a preset period, the local register latches the stable and effective feeding flow rate under the current operating conditions, the dynamic pre-truncation threshold is kept updated in real time to match the channel status, and the edge management gateway maintains basic wireless communication links with the execution terminal and the field sensing terminal.

[0048] The edge control gateway continuously monitors the CQI value of the downlink wireless link. When the CQI is less than the preset deep fading threshold, it determines that the wireless link has entered a deep fading state, and the control signaling transmission delay exceeds the predictable range. The preset deep fading threshold is set to 3. This value corresponds to a modulation and coding scheme of quadrature phase shift keying, a code rate below 0.2, a channel bit error rate above 10 to the power of -2, and a significant increase in retransmissions, making it impossible to ensure batching accuracy through conventional threshold adjustments. When CQI=0 is detected, it is directly determined that the downlink is completely interrupted, and the edge control gateway stops sending control signals. The execution terminal relies solely on the timeout mechanism of not receiving signals for 50 consecutive milliseconds to automatically trigger local shutdown as a fallback. When CQI is greater than 0 and less than the preset deep fading threshold, the edge control gateway immediately sends a physical throttling signal to the execution terminal, driving the execution terminal to adjust the batching valve to a preset opening. The preset opening is set to 10%, which can reduce the feeding flow rate to one-tenth of the original flow rate, significantly reducing the overshoot risk caused by delay fluctuations.

[0049] While the edge control gateway issues a physical throttling signal, it also starts a response masking timer to freeze the reset operation of the dynamic pre-truncation threshold. The preset duration of the response masking timer is set to 20 milliseconds, which covers the maximum response time of 15 milliseconds for an industrial proportional servo valve from fully open to 10% opening, with a 5-millisecond margin for action. The response masking timer is triggered only once after it starts. During operation, it continuously monitors the channel status. If it is detected that the CQI value obtained in each of the three consecutive reporting cycles is greater than the preset recovery threshold, the timer is immediately terminated, the threshold freeze is lifted, and the throttling recovery process is executed. If the CQI is detected to be greater than 0 and less than the preset deep fading threshold again during the timer's operation, the timer will not be triggered again.

[0050] After the response shielding timer reaches its preset duration, the edge control gateway replaces the real-time feeding flow rate with a preset constant flow rate, recalculates the overshoot compensation amount, and completes the reset of the dynamic pre-truncation threshold. The constant flow rate is determined based on the standard feeding flow rate corresponding to the preset opening degree of the batching valve and has a linear positive correlation with the valve opening degree. Multiple sets of corresponding constant flow rates can be preset according to different valve opening degrees. In this embodiment, when the valve opening degree is 10%, the corresponding constant flow rate is 5 kg / s.

[0051] After the edge control gateway issues a physical throttling signal, it continuously monitors the channel state recovery. When the CQI value acquired in each of the three consecutive reporting cycles is greater than the preset recovery threshold, the channel state is considered to have stabilized. The preset recovery threshold is set to 7, corresponding to hexadecimal quadrature amplitude modulation (QAM), a code rate higher than 0.45, and a channel bit error rate lower than 10^-5, meeting the reliability and latency requirements of ultra-reliable low-latency communication services. The edge control gateway then issues a throttling release signal to the execution terminal, causing the batching valve to return to its original operating opening. Simultaneously, it replaces the constant flow rate with the real-time feed rate, recalculates the overshoot compensation, and restores the normal update logic of the dynamic pre-truncation threshold, allowing the system to return to its normal operating control flow.

[0052] The edge control gateway synchronously monitors the reception status of weight messages. When the duration of no weight message reception reaches a preset disconnection span threshold, it determines that the uplink sensing link between the field sensing terminal and the edge control gateway has been continuously interrupted. The preset disconnection span threshold is set to 3 weight message reporting cycles, corresponding to a duration of 3 milliseconds. The basis for setting this threshold is that if no message is received within this duration, the uplink has been continuously interrupted, and it is impossible to rely on real-time weight messages for threshold comparison. At the same time, the maximum discharge rate of aluminum alloy liquid material is 50 kg / s, and the weight change within 3 milliseconds does not exceed 0.15 kg, which is close to the maximum allowable error range for industrial batching. The edge control gateway calculates the difference between the current system time and the calibration timestamp corresponding to the most recently received weight message as the time domain increment. It multiplies the time domain increment by the discharge rate extracted from the local register, and adds the weight value contained in the weight message to obtain the virtual inferred weight. The physical dimension of the time-domain increment is seconds, the physical dimension of the feed flow rate is kilograms per second, and the physical dimension of the weight increment obtained by multiplying the two is kilograms, which is completely consistent with the physical dimension of the measured weight.

[0053] The edge control gateway continuously compares the virtual inferred weight with the dynamic pre-truncation threshold. If a valid weight message reported by the field sensing terminal is received again before the pre-truncation signaling is issued, the uplink is determined to have recovered. The self-healing fallback logic is immediately exited, and the asynchronous threshold comparison process under normal operating conditions is switched to. At the same time, a 5-millisecond state-holding timer is started. During the timer's operation, even if no weight message is received, the self-healing fallback logic is not re-triggered. Only if no weight message is received after the timer expires will the disconnection determination be restarted. When the virtual inferred weight reaches the dynamic pre-truncation threshold, a pre-truncation signaling is immediately issued to the execution terminal. The issuance of this signaling also follows the signaling interlock logic to complete the forced truncation operation.

[0054] The execution terminal has built-in local fallback control logic, which pre-stores the preset target threshold for this batching operation in its local storage unit. During normal batching, the edge control gateway continuously sends heartbeat keep-alive signals to the execution terminal at a preset period (e.g., 1 millisecond) that is consistent with the channel state parameter comparison period of the gateway. It also generates a real air interface retransmission count value in real time through a high-frequency downlink probe data stream for extraction and calculation, and resets the signaling reception timer of the execution terminal. If no control signaling or heartbeat keep-alive signaling is received from the edge control gateway for 50 consecutive milliseconds, and the batching weight corresponding to the accumulated batching time in the local timing unit has reached 95% of the preset target threshold, a hardware interrupt is automatically triggered to shut down the batching valve. After communication is restored, a local truncation event is reported to the edge control gateway.

[0055] After this step is completed, the system will have completed adaptive control and safety backup under extreme abnormal conditions, achieving precise control of aluminum alloy material batching under all operating conditions.

[0056] This embodiment fully realizes cross-layer collaborative compensation for communication layer latency and physical layer overshoot during aluminum alloy batching. This solution ensures a unified time base for the entire system through PTP master-slave clock synchronization, providing a precise time scale for flow rate measurement and latency calculation; it achieves stable extraction of the feed flow rate through sliding window smoothing filtering, eliminating detection noise caused by molten pool sloshing; it transforms underlying communication parameters into advance parameters for physical control through RNIS cross-layer air interface status perception, fundamentally offsetting the valve closing lag problem caused by transmission latency; it avoids control deadlock caused by message reception interruption through asynchronous threshold detection and high-priority signaling scheduling, ensuring real-time and reliable execution of truncation commands; and it achieves a safety fallback in extreme scenarios through adaptive management and self-healing logic for abnormal operating conditions. In industrial wireless environments with strong electromagnetic interference, this solution can control the batching error caused by network nondeterministic latency within ±0.1kg. Compared with traditional fixed threshold batching systems, the batching accuracy is significantly improved, greatly enhancing the pass rate of aluminum alloy composition control and the consistency of product performance.

[0057] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. A system for optimizing and adjusting the composition of aluminum alloy feedstock, characterized in that: This includes on-site sensing terminals, edge control gateways, and execution terminals; The field sensing terminal reports weighted messages with timestamps to the edge control gateway at a preset period. The edge control gateway extracts the unloading flow rate based on the continuously received weight messages, and obtains the channel quality indicator and retransmission count value of the wireless link where the execution terminal is located. The edge control gateway calculates the estimated dwell time based on the channel quality indicator and the retransmission count value, and calculates the overshoot compensation amount based on the estimated dwell time and the material flow rate, so as to reset the preset target threshold to the dynamic pre-truncation threshold. The edge control gateway has a built-in asynchronous threshold comparator, which calculates the current transient weight by combining the received weight message with the material flow rate. When the weight contained in the weight message or the transient weight reaches the dynamic pre-truncation threshold, the edge management gateway sends a pre-truncation signaling carrying a priority flag to the execution terminal. The execution terminal triggers a hardware interrupt based on the pre-truncation signaling to drive the batching valve to close.

2. The aluminum alloy batching optimization and composition adjustment system according to claim 1, characterized in that: The edge control gateway includes a flow rate observation module and a local register; The flow velocity observation module extracts a sequence of messages that meet a preset duration threshold from the continuously received weight messages based on a sliding time window. The flow rate observation module performs smoothing filtering on the message sequence to extract the weight trend component, and calculates the feeding flow rate based on the weight trend component; The flow rate observation module writes the feeding flow rate into the local register, which is then read by the edge control gateway to calculate the overshoot compensation amount.

3. The aluminum alloy batching optimization and composition adjustment system according to claim 2, characterized in that: The field sensing terminal interacts with the edge control gateway to perform master-slave clock synchronization in order to extract the clock offset. The edge control gateway uses the clock offset to compensate for the timestamp carried in the weight message to generate a calibration timestamp; The flow velocity observation module defines the true span of the sliding time window based on the extracted calibration timestamp, and uses this as a comparison benchmark to determine whether the message sequence meets the preset duration threshold.

4. The aluminum alloy batching optimization and composition adjustment system according to claim 1, characterized in that: The edge management gateway has a pre-set mapping table for basic air interface frame length, single retransmission time and backoff compensation. The edge management gateway obtains the channel quality indicator and the retransmission count value through the wireless network information service interface; The edge management gateway retrieves the queue backoff compensation amount from the backoff compensation mapping table using the channel quality indicator as an index, and sums the product of the basic air interface frame length, the retransmission count value and the single retransmission time, and the queue backoff compensation amount to calculate the estimated dwell time.

5. The aluminum alloy batching optimization and composition adjustment system according to claim 4, characterized in that: When the retransmission count value is greater than a preset reference threshold or the drop value of the channel quality indicator in an adjacent period reaches a preset attenuation order, the edge management gateway triggers the calculation of the estimated dwell time. The edge control gateway multiplies the estimated dwell time by the material flow rate to obtain the overshoot compensation amount, and subtracts the overshoot compensation amount from the preset target threshold to reset the dynamic pre-truncation threshold.

6. The aluminum alloy batching optimization and composition adjustment system according to claim 5, characterized in that: The asynchronous threshold comparator operates independently of the reception cycle of the weight message. The time difference between the current system time and the timestamp carried in the most recently received weight message is calculated. The transient weight is obtained by linear extrapolating the product of the time difference and the feed flow rate, plus the weight contained in the weight message.

7. The aluminum alloy batching optimization and composition adjustment system according to claim 6, characterized in that: The execution terminal includes a communication module, an internal bus, and a programmable logic controller; The edge management gateway configures the network layer header of the pre-truncation signaling with a preset priority to form the priority tag; The communication module parses the pre-truncation signaling carrying the priority flag and triggers the hardware interrupt to the programmable logic controller via the internal bus; The programmable logic controller responds to the hardware interrupt to shut off the dispensing valve.

8. The aluminum alloy batching optimization and composition adjustment system according to claim 5, characterized in that: The edge control gateway has a response shielding timer and a constant flow rate; When the channel quality indicator is less than the preset deep fading threshold, the edge control gateway sends a physical throttling signal to drive the execution terminal to adjust the batching valve to the preset opening degree and start the response shielding timer to freeze the reset of the dynamic pre-truncation threshold; After the response shielding timer reaches a preset duration, the edge control gateway recalculates the overshoot compensation amount by replacing the feed flow rate with the constant flow rate, thereby resetting the dynamic pre-truncation threshold.

9. The aluminum alloy batching optimization and composition adjustment system according to claim 8, characterized in that: After the edge control gateway issues the physical throttling signaling, if the number of times the channel quality indication obtained continuously exceeds the preset recovery threshold reaches the preset stable period number, then the throttling release signaling is issued to the execution terminal, and the overshoot compensation amount is recalculated by replacing the constant flow rate with the feeding flow rate.

10. The aluminum alloy batching optimization and composition adjustment system according to claim 3, characterized in that: When the edge control gateway fails to receive the weight message for a duration that reaches a preset disconnection span threshold, it calculates the difference between the current system time and the calibration timestamp corresponding to the most recently received weight message as a time domain increment. The time domain increment is then multiplied by the material flow rate extracted from the local register, and the weight contained in the weight message is added to obtain a virtual inferred weight. When the virtual inferred weight reaches the dynamic pre-truncation threshold, the pre-truncation signaling is sent to the execution terminal.

Images