Distributed manufacturing multi-source data collaboration and closed-loop control method driven by process tact time

Abstract

The application discloses a kind of process beat driving distributed manufacturing multi-source data cooperation and closed-loop control method, belong to industrial automation and intelligent manufacturing field, through real-time acquisition main drive motor operating frequency by sensor and calculate real-time capacity, anchor the main control beat of whole line cooperation;Based on front-end discharge bin level and change rate, a cross-process load prediction model is constructed, and the operating beat of the feed frequency converter is dynamically adjusted;Using industrial communication network to summarize multi-process operation parameters to establish interaction matrix, extract the load feature vector of the back-end process for real-time state monitoring;When triggering early warning, execute adaptive speed reduction algorithm to fine-tune the main drive frequency, and simultaneously adjust the heating power compensation of each temperature zone;Under the condition of maintaining the continuous operation of the system, the abnormal non-stop closed-loop disposal is realized, the consistency of the material sintering thermal history is ensured, the problem of whole line shutdown and material scrap caused by local fluctuation is effectively solved, and the production stability and product quality consistency of distributed manufacturing are improved.

Description

Technical Field

[0001] This application belongs to the field of industrial automation and intelligent manufacturing, specifically involving a distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time. Background Technology

[0002] With the in-depth development of intelligent manufacturing technology, distributed manufacturing systems have been widely applied in the field of new energy material processing, becoming a key path to improve production flexibility and resource allocation efficiency. Traditional manufacturing processes rely on automated control logic to achieve orderly connection of multiple links, which is of significant value in ensuring the overall continuity of production line operations and improving capacity utilization. In complex material synthesis scenarios, the stability of the production cycle and the accuracy of data interaction between processes directly determine the consistency of materials in physical and chemical reaction processes, thus affecting the performance and quality of the final product.

[0003] Among them, roller kiln sintering is a core link in the production of materials such as lithium iron phosphate, and its operating line speed constitutes the core driving pulse of the entire production line. This process typically receives the supply from the upstream spray drying and outputs it to the downstream pulverization process, forming a highly coupled chain-like operating environment. Due to the complex physicochemical reactions involved in the sintering process and its extremely high thermal inertia, the system places extremely high technical requirements on the synchronization of cycle times between processes, real-time capacity mapping, and dynamic prediction of cross-process loads.

[0004] However, traditional control methods struggle to accurately predict cross-process loads, leading to logical lags in data coordination between upstream material supply and downstream output, impacting the overall flow balance of the production line. Furthermore, existing systems lack closed-loop handling capabilities for sudden anomalies. When overload or blockage warnings occur in the downstream crushing process, emergency shutdowns are often forced due to the inability to dynamically and finely adjust the core process cycle time, risking the scrapping of several tons of material in the high-temperature kiln. In addition, traditional linear adjustment logic cannot effectively address nonlinear parameter fluctuations during sintering, making it difficult to actively compensate for furnace power while adjusting the production cycle time, resulting in unstable product sintering quality during dynamic adjustments.

[0005] Therefore, a distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time is desired. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to provide a distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time, which addresses the shortcomings of the prior art.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: A distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time includes the following steps: Step 1, anchoring the main control cycle of the entire distributed manufacturing line: the operating frequency of the main drive roller is collected in real time by the sensor installed at the end of the main drive motor shaft of the roller kiln. The operating frequency is used as the heartbeat signal for the entire line to coordinate. Based on the structural parameters, transmission ratio and real-time frequency of the main drive roller, the current operating line speed is calculated and then mapped to the real-time production capacity output per unit time. The production capacity output is measured in a preset mass unit. The specific formula for calculating the linear velocity is as follows: ; in, This represents the corrected output linear velocity of the main drive roller. This represents the initial linear velocity set based on the baseline capacity. This represents the moisture correction factor. This represents the measured moisture content of the material. Represents the standard moisture content threshold. Represents the target production capacity constant. This represents the currently measured load per unit length. Step 2, construct a cross-process load prediction model: establish a chain collaborative model covering sand mill inventory, spray power, roller speed and air crushing frequency. By real-time monitoring of the material level height and material level change rate of the unloading bin of the front spray drying tower, predict the material flow demand of the subsequent process, and dynamically adjust the frequency converter operation rhythm of the roller kiln feed end according to the prediction results, so that the unloading frequency and the front feeding speed are logically matched. Step 3, Implement multi-source data collaboration and status monitoring: Real-time collection of operating parameters of sand milling, spraying, sintering and air crushing processes through industrial communication network, establish multi-source data interaction matrix, focus on monitoring the main motor current value, feed valve opening and cyclone separator pressure difference of the back-end airflow crushing process, and perform dimensionality reduction processing on the interaction matrix through principal component analysis algorithm to extract feature vectors representing load status; Step 4, Abnormal Closed-Loop Handling and Adaptive Adjustment: When the characteristic vector of the downstream airflow pulverizing process triggers the current overload or the material blockage warning threshold, the control system, while maintaining the continuous operation of the roller kiln, starts the adaptive deceleration algorithm to fine-tune the frequency of the main drive roller. At the same time, according to the deceleration ratio, the power compensation amount of each heating temperature zone inside the kiln is synchronously adjusted through the heat balance equation. Under the premise of ensuring the consistency of the sintering heat process of the material, the non-stop closed-loop collaborative control of the entire production line is realized.

[0008] The specific power compensation calculation formula is as follows: ; in, This represents the adjusted heating power. This represents the rated heating power for the corresponding temperature zone. Represents the current actual linear velocity. Represents the rated linear velocity. Represents the heat exchange influencing factor. This represents the deviation in temperature field prediction. This represents the thermal response time constant of the kiln.

[0009] Preferably, the process of anchoring the main control cycle of the entire line in step 1 includes: using an incremental photoelectric encoder with a predetermined resolution to monitor the roller rotation speed, and setting the sampling period to a preset sampling time interval; the controller receives the phase differential pulse signal output by the incremental photoelectric encoder through a counting module, and extracts the real-time angular velocity of the motor rotor using frequency multiplication processing logic; the controller performs frequency-voltage conversion processing on the collected pulse signal to obtain an analog quantity reflecting the physical linear velocity; the linear velocity is multiplied by the loading capacity of a single tray and the distribution density of trays per unit length to obtain the real-time material flow rate; the material flow rate is used as a global reference reference signal and distributed to the front-end metering pump controller and the back-end screw feeder controller through an industrial network based on the real-time Ethernet protocol. After receiving the reference signal, each controller performs local closed-loop adjustment through a preset proportional coefficient to ensure that each physical unit in the distributed system is synchronized on the time axis.

[0010] Preferably, the process of establishing the cross-process load prediction model in step 2 includes: using multi-sensor fusion technology to monitor the material level in the unloading hopper of the spray drying tower; installing multiple sets of frequency-modulated continuous wave radar material level gauges at the top of the unloading hopper to measure the spatial distance on the material surface; and installing multiple sets of pressure strain sensors in the conical hopper of the unloading hopper to sense the gravity distribution of the material; using a weighted average algorithm to fuse multiple sets of radar signals and pressure signals to filter out measurement errors caused by uneven material accumulation, hanging material, or dust; the model calculates the material level height value and its derivative with respect to time in real time to obtain the material level change rate; when the material level in the unloading hopper is detected to exceed the preset high-level warning threshold, the system sends an acceleration command to the roller kiln feed frequency converter through the fieldbus to increase the discharging frequency by a preset adjustment range to accelerate material digestion; when the material level in the unloading hopper is detected to be lower than the preset low-level threshold, the discharging frequency is simultaneously reduced; The cross-process load prediction model uses the following formula to dynamically predict the material flow requirements of subsequent processes: ; in, Represents prediction of the future Material demand flow after time, This represents the material inflow rate corresponding to the current feeding cycle. This represents the rate of change in material level in the unloading hopper. This represents the current material level in the unloading hopper. This represents the target material level setting value. This represents the current operating power of the spray drying tower. , , These are weighting coefficients obtained through experience or identification; Preferably, the chain-coordinated model involves a feedforward correction logic for the material moisture content, including: acquiring real-time moisture data of the sprayed material by installing an online infrared moisture meter at the outlet of the spray drying tower; when the moisture content of the material represented by the moisture data exceeds a preset moisture standard value, the chain-coordinated model automatically corrects the capacity output mapping coefficient by reducing the roller speed to extend the dehydration time of the material in the kiln; the feedforward correction logic calculates the corrected output linear speed of the main drive roller based on the moisture deviation value, the baseline speed, the moisture correction coefficient, and the target capacity constant, realizing dynamic speed adjustment based on the material properties, and ensuring that the material received by the downstream airflow pulverizing process is in a fragile and dry state.

[0011] Preferably, the industrial communication network involved in the multi-source data collaboration in step 3 adopts the real-time Ethernet protocol, and the data synchronization cycle is within a preset cycle range; the rows of the multi-source data interaction matrix represent each physical process node, and the columns represent the dimensions of pressure, temperature, current, frequency, flow rate, and vibration parameters; the system extracts the key principal components that contribute more than a preset proportion to the system stability by performing eigenvalue decomposition on the covariance matrix and defines them as the feature vector; for the back-end airflow pulverization process, the system collects the pulverizing disc speed, classifying wheel frequency, active power of the main drive motor, three-phase current value, and opening feedback of the feed rotary valve in real time, and realizes the early detection of material blockage tendency by establishing a correlation function between current fluctuation rate and material bulk density; The following multi-source data fusion formula is used to quantify the load status of the back-end processes: ; in, This represents the overall load index of the back-end processes. This represents the current of the main motor for airflow pulverization. Represents the rated current. This represents the pressure difference between the inlet and outlet of the cyclone separator. Represents the maximum permissible differential pressure. This represents the vibration amplitude of the crushing disc. This represents the normal vibration reference value. , , These are the weighting coefficients, and their sum is 1.

[0012] Preferably, the logic of the adaptive speed reduction algorithm in step 4 includes: when the main motor current of the air jet milling process reaches a preset current threshold and the duration exceeds a preset time threshold, the system determines that there is a slight risk of overload; the controller gradually reduces the main drive frequency of the roller kiln according to a preset frequency gradient through a graded voltage reduction frequency adjustment strategy until it reaches the preset safe frequency lower limit; at the same time, the system reduces the rotation speed of the feed feeder and reduces the density of the load entering the kiln from the source by reverse adjustment of the front-end feed inverter; the abnormal closed-loop handling also includes self-recovery logic based on the current recovery rate, when the main motor current of the back-end air jet milling process falls below the preset safe value and the fluctuation rate tends to stabilize, the system starts a step-by-step speed increase scheme; The graded voltage reduction frequency adjustment strategy uses the following dynamic step size formula to determine the rate reduction amount at each step: ; in, This represents the frequency change of the next deceleration step. Represents the baseline deceleration step size. This represents the current current of the main motor for airflow pulverization. Represents the overload warning threshold current. Represents the rate of change of current. , A positive adjustment coefficient; Furthermore, the stepped acceleration scheme uses the following smooth curve formula to set the target frequency during the recovery process: ; in, Represents the target frequency at the current moment. This represents the lower safety limit frequency during anomaly handling. Represents the target recovery frequency. This represents the acceleration time constant.

[0013] Preferably, the power compensation adjustment and speed reduction algorithm execute strongly coupled control logic, including: the system calculates the power reduction coefficient corresponding to each temperature zone in real time according to the preset ratio of the current speed reduction through a preset heat balance equation; when the roller speed decreases by a preset amount, the set temperature of the high temperature zone is slightly adjusted downward by the preset temperature deviation, or the heating power is reduced synchronously by adjusting the duty cycle of the thyristor of the heating control component; the heat balance equation performs power compensation calculation based on the current actual linear speed, rated linear speed, heat exchange influence factor, temperature field prediction deviation, and the thermal response time constant of the kiln, to ensure that the total energy absorbed by the material under the variable cycle operation state is consistent with the standard process curve, and to eliminate the temperature field deviation caused by speed changes; A power compensation correction based on thermal history consistency is introduced, specifically: the cumulative heat absorbed by the material in the kiln is calculated in real time, and the heating power is additionally corrected according to the deviation from the standard heat. The correction formula is as follows: ; ; in, This represents the cumulative heat absorbed by the material inside the kiln. , These represent the times when materials enter and leave the kiln, respectively. Represents real-time heat exchange efficiency. This represents the target heat that the material should absorb under standard processes. This represents the power compensation correction amount. Represents the thermal inertia compensation coefficient. This represents the rated power for the current temperature range.

[0014] Preferably, the method integrates a digital twin monitoring system, including: constructing a virtual mirror of the entire production line using a 3D modeling engine on a host computer interface, and rendering the flow status of materials across the entire line in real time; the system renders the movement trajectory of each tray in the virtual space based on the real-time calculated line velocity and predicts the time to reach the downstream process; the system marks the load status of each process on the 3D interface with different colors, where the first color represents normal synchronization of the entire line's cycle time, the second color represents that the predicted load of the downstream process is too high and has triggered the feed current limit, and the third color represents that it has entered a closed-loop handling state of adaptive speed reduction and power compensation; the digital twin monitoring system also integrates an augmented reality interface to overlay real-time cycle time data, current waveforms, and material position information on top of the physical equipment.

[0015] Preferably, the method includes a model predictive control strategy for long-cycle thermal inertia: the controller, based on a first-order transfer function model with pure time delay, predicts the furnace temperature field changes within a future preset time period according to the temperature fluctuation trend within a preset historical time period and the current roller speed adjustment; if the prediction result deviates from the process curve, the system pre-sets an intervention time to adjust the speed of the circulating fan or the set point of the heating group, using advanced control to offset the temperature overshoot or lag caused by thermal inertia; the system also receives meteorological data from the factory location, including atmospheric pressure, humidity and outdoor temperature, and uses environmental parameters as feedforward compensation to fine-tune the bias value of the heating power in advance, ensuring the robustness of the production line cycle under different climatic conditions.

[0016] Preferably, the method employs a high-precision positioning and calibration system in the material conveying stage: high-temperature resistant passive radio frequency identification (RFID) readers are installed at the feed inlet and discharge outlet of the roller kiln, and a unique electronic tag is embedded in the bottom of each container; the system records the entry time, exit time, and timestamp of each container passing through the junction of each heating section in real time; the timestamp data is cross-checked with the theoretical position calculated by the main control cycle; if the deviation between the physical position and the logically calculated position exceeds a preset deviation threshold, the system automatically triggers the cycle calibration logic to correct the speed mapping coefficient; when the downstream process encounters an anomaly and requires low-speed operation, the system feeds back the signal to the upstream sand milling process, and the sand mill control system automatically reduces the slurry circulation flow rate according to the buffer capacity of the spray drying tower.

[0017] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects: 0. By anchoring the main drive roller speed frequency as the heartbeat of the entire line, this invention achieves a high degree of synchronization between distributed processes. For process links with large thermal inertia that cannot be easily stopped, the abnormal closed-loop handling scheme proposed in this invention can mitigate risks by adaptively reducing speed rather than emergency shutdown when a material blockage warning occurs at the back end. This breakthrough avoids large-scale kiln shutdown failures caused by minor disturbances at the back end, effectively solves the industry problem of high-temperature materials facing scrapping mentioned in the background technology, and significantly reduces the unplanned downtime rate of the production line. 0. The chain-based collaborative model established by this invention breaks the data silos between processes in traditional manufacturing. By monitoring multi-dimensional data such as material level in the front-end unloading hopper, spray power, and material moisture, the system can predict the load-bearing pressure of subsequent processes in real time. The logic of dynamically adjusting the cycle time of the feed inverter ensures that the material supply and the capacity output of the core process are always in a dynamic balance. Compared with the traditional linear adjustment logic, this invention significantly reduces the redundancy of material accumulation inside the production line, greatly improving the flow coordination efficiency and capacity utilization of the entire line. 0. To address the issue of thermal history fluctuations caused by frequency adjustment, this invention introduces a synchronous power compensation mechanism. By actively compensating for furnace power while adjusting the frequency, this invention eliminates temperature field deviations caused by speed changes, ensuring a constant energy absorption of materials during physical and chemical reactions. This feedforward compensation and collaborative control logic keeps the fluctuation rate of physical performance indicators of the product at an extremely low level under system fluctuation conditions, ensuring high-quality output in a large-scale distributed manufacturing environment.

[0018] 0. The combination of multi-source data interaction matrix and adaptive adjustment algorithm enables the system to make autonomous decisions in the face of sudden anomalies. From the extraction of current feature vectors to the execution of graded deceleration strategy, and then to the self-recovery acceleration after current recovery, the entire process does not require manual intervention and the response speed reaches the millisecond level. This intelligent closed-loop control method not only reduces the dependence on human experience, but also provides solid technical support for the subsequent construction of fully automated and unmanned smart factories. Attached Figure Description

[0019] Figure 1 This is a flowchart of the distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time of the present invention; Figure 2 This is a schematic diagram of the principle framework of the process cycle time-driven and multi-source data collaboration in the distributed manufacturing multi-source data collaboration and closed-loop control method of the present invention. Figure 3 This is a logical flow diagram of the whole-line master control cycle anchoring and distributed physical unit synchronous control in the distributed manufacturing multi-source data collaboration and closed-loop control method driven by the process cycle of the present invention. Figure 4 This is a schematic diagram of a chain-like collaborative model for cross-process load prediction and dynamic matching of material flow in the distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time of the present invention. Figure 5 This is a data flow diagram illustrating the construction of the multi-source data interaction matrix and extraction of key load features in the distributed manufacturing multi-source data collaboration and closed-loop control method driven by the process cycle of the present invention. Figure 6 This is a flowchart illustrating the closed-loop control process of adaptive deceleration and heating power compensation coupling adjustment under abnormal conditions in the distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention. The present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments. The purpose and effects of the present invention will become clearer. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention.

[0021] like Figures 1 to 6 As shown, a distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time is used, such as... Figure 1 As shown, it includes the following steps: Step 1, anchoring the main control cycle of the entire distributed manufacturing line: the operating frequency of the main drive roller is collected in real time by the sensor installed at the end of the main drive motor shaft of the roller kiln. The operating frequency is used as the heartbeat signal for the entire line to coordinate. Based on the structural parameters, transmission ratio and real-time frequency of the main drive roller, the current operating line speed is calculated and then mapped to the real-time production capacity output per unit time. The production capacity output is measured in a preset mass unit. The specific formula for calculating the linear velocity is as follows: ; in, This represents the corrected output linear velocity of the main drive roller. This represents the initial linear velocity set based on the baseline capacity. This represents the moisture correction factor. This represents the measured moisture content of the material. Represents the standard moisture content threshold. Represents the target production capacity constant. This represents the currently measured load per unit length. Step 2: Construct a cross-process load prediction model: Establish a chain-like collaborative model covering sand mill inventory, spray power, roller speed, and air crushing frequency. By real-time monitoring of the material level height and material level change rate of the unloading bin of the front-end spray drying tower, predict the material flow demand of subsequent processes, and dynamically adjust the operating rhythm of the frequency converter at the feed end of the roller kiln according to the prediction results, so that the unloading frequency and the front-end feeding speed are logically matched. Step 3: Implement multi-source data collaboration and status monitoring: Real-time collection of operating parameters of sand milling, spraying, sintering, and air crushing processes through industrial communication networks, establish a multi-source data interaction matrix, focus on monitoring the main motor current value, feed valve opening, and cyclone separator pressure difference of the back-end airflow pulverizing process, and perform dimensionality reduction processing on the interaction matrix through principal component analysis algorithm to extract feature vectors representing the load status. Step 4, Abnormal Closed-Loop Handling and Adaptive Adjustment: When the characteristic vector of the downstream airflow pulverizing process triggers the current overload or the material blockage warning threshold, the control system, while maintaining the continuous operation of the roller kiln, starts the adaptive deceleration algorithm to fine-tune the frequency of the main drive roller. At the same time, according to the deceleration ratio, the power compensation amount of each heating temperature zone inside the kiln is synchronously adjusted through the heat balance equation. Under the premise of ensuring the consistency of the sintering heat process of the material, the non-stop closed-loop collaborative control of the entire production line is realized.

[0022] The specific power compensation calculation formula is as follows: ; in, This represents the adjusted heating power. This represents the rated heating power for the corresponding temperature zone. Represents the current actual linear velocity. Represents the rated linear velocity. Represents the heat exchange influencing factor. This represents the deviation in temperature field prediction. This represents the thermal response time constant of the kiln.

[0023] Preferably, the process of anchoring the main control cycle of the entire line in step 1 includes: using an incremental photoelectric encoder with a predetermined resolution to monitor the roller rotation speed, and setting the sampling period to a preset sampling time interval; the controller receives the phase differential pulse signal output by the incremental photoelectric encoder through a counting module, and extracts the real-time angular velocity of the motor rotor using frequency multiplication processing logic; the controller performs frequency-voltage conversion processing on the collected pulse signal to obtain an analog quantity reflecting the physical linear velocity; the linear velocity is multiplied by the loading capacity of a single tray and the distribution density of trays per unit length to obtain the real-time material flow rate; the material flow rate is used as a global reference reference signal and distributed to the front-end metering pump controller and the back-end screw feeder controller through an industrial network based on the real-time Ethernet protocol. After receiving the reference signal, each controller performs local closed-loop adjustment through a preset proportional coefficient to ensure that each physical unit in the distributed system is synchronized on the time axis.

[0024] Preferably, the process of establishing the cross-process load prediction model in step 2 includes: using multi-sensor fusion technology to monitor the material level in the unloading hopper of the spray drying tower; installing multiple sets of frequency-modulated continuous wave radar material level gauges at the top of the unloading hopper to measure the spatial distance on the material surface; and installing multiple sets of pressure strain sensors in the conical hopper of the unloading hopper to sense the gravity distribution of the material; using a weighted average algorithm to fuse multiple sets of radar signals and pressure signals to filter out measurement errors caused by uneven material accumulation, hanging material, or dust; the model calculates the material level height value and its derivative with respect to time in real time to obtain the material level change rate; when the material level in the unloading hopper is detected to exceed the preset high-level warning threshold, the system sends an acceleration command to the roller kiln feed frequency converter through the fieldbus to increase the discharging frequency by a preset adjustment range to accelerate material digestion; when the material level in the unloading hopper is detected to be lower than the preset low-level threshold, the discharging frequency is simultaneously reduced; The cross-process load prediction model uses the following formula to dynamically predict the material flow requirements of subsequent processes: ; in, Represents prediction of the future Material demand flow after time, This represents the material inflow rate corresponding to the current feeding cycle. This represents the rate of change in material level in the unloading hopper. This represents the current material level in the unloading hopper. This represents the target material level setting value. This represents the current operating power of the spray drying tower. , , These are weighting coefficients obtained through experience or identification; Preferably, the chain-coordinated model involves a feedforward correction logic for the material moisture content, including: acquiring real-time moisture data of the sprayed material by installing an online infrared moisture meter at the outlet of the spray drying tower; when the moisture content of the material represented by the moisture data exceeds a preset moisture standard value, the chain-coordinated model automatically corrects the capacity output mapping coefficient by reducing the roller speed to extend the dehydration time of the material in the kiln; the feedforward correction logic calculates the corrected output linear speed of the main drive roller based on the moisture deviation value, the baseline speed, the moisture correction coefficient, and the target capacity constant, realizing dynamic speed adjustment based on the material properties, and ensuring that the material received by the downstream airflow pulverizing process is in a fragile and dry state.

[0025] Preferably, the industrial communication network involved in the multi-source data collaboration in step 3 adopts the real-time Ethernet protocol, and the data synchronization cycle is within a preset cycle range; the rows of the multi-source data interaction matrix represent each physical process node, and the columns represent the dimensions of pressure, temperature, current, frequency, flow rate, and vibration parameters; the system extracts the key principal components that contribute more than a preset proportion to the system stability by performing eigenvalue decomposition on the covariance matrix and defines them as the feature vector; for the back-end airflow pulverization process, the system collects the pulverizing disc speed, classifying wheel frequency, active power of the main drive motor, three-phase current value, and opening feedback of the feed rotary valve in real time, and realizes the early detection of material blockage tendency by establishing a correlation function between current fluctuation rate and material bulk density; The following multi-source data fusion formula is used to quantify the load status of the back-end processes: ; in, This represents the overall load index of the back-end processes. This represents the current of the main motor for airflow pulverization. Represents the rated current. This represents the pressure difference between the inlet and outlet of the cyclone separator. Represents the maximum permissible differential pressure. This represents the vibration amplitude of the crushing disc. This represents the normal vibration reference value. , , These are the weighting coefficients, and their sum is 1.

[0026] Preferably, the logic of the adaptive speed reduction algorithm in step 4 includes: when the main motor current of the air jet milling process reaches a preset current threshold and the duration exceeds a preset time threshold, the system determines that there is a slight risk of overload; the controller gradually reduces the main drive frequency of the roller kiln according to a preset frequency gradient through a graded voltage reduction frequency adjustment strategy until it reaches the preset safe frequency lower limit; at the same time, the system reduces the rotation speed of the feed feeder and reduces the density of the load entering the kiln from the source by reverse adjustment of the front-end feed inverter; the abnormal closed-loop handling also includes self-recovery logic based on the current recovery rate, when the main motor current of the back-end air jet milling process falls below the preset safe value and the fluctuation rate tends to stabilize, the system starts a step-by-step speed increase scheme; The graded voltage reduction frequency adjustment strategy uses the following dynamic step size formula to determine the rate reduction amount at each step: ; in, This represents the frequency change of the next deceleration step. Represents the baseline deceleration step size. This represents the current current of the main motor for airflow pulverization. Represents the overload warning threshold current. Represents the rate of change of current. , A positive adjustment coefficient; Furthermore, the stepped acceleration scheme uses the following smooth curve formula to set the target frequency during the recovery process: ; in, Represents the target frequency at the current moment. This represents the lower safety limit frequency during anomaly handling. Represents the target recovery frequency. This represents the acceleration time constant.

[0027] Preferably, the power compensation adjustment and speed reduction algorithm execute strongly coupled control logic, including: the system calculates the power reduction coefficient corresponding to each temperature zone in real time according to the preset ratio of the current speed reduction through a preset heat balance equation; when the roller speed decreases by a preset amount, the set temperature of the high temperature zone is slightly adjusted downward by the preset temperature deviation, or the heating power is reduced synchronously by adjusting the duty cycle of the thyristor of the heating control component; the heat balance equation performs power compensation calculation based on the current actual linear speed, rated linear speed, heat exchange influence factor, temperature field prediction deviation, and the thermal response time constant of the kiln, to ensure that the total energy absorbed by the material under the variable cycle operation state is consistent with the standard process curve, and to eliminate the temperature field deviation caused by speed changes; A power compensation correction based on thermal history consistency is introduced, specifically: the cumulative heat absorbed by the material in the kiln is calculated in real time, and the heating power is additionally corrected according to the deviation from the standard heat. The correction formula is as follows: ; ; in, This represents the cumulative heat absorbed by the material inside the kiln. , These represent the times when materials enter and leave the kiln, respectively. Represents real-time heat exchange efficiency. This represents the target heat that the material should absorb under standard processes. This represents the power compensation correction amount. Represents the thermal inertia compensation coefficient. This represents the rated power for the current temperature range.

[0028] Preferably, the method integrates a digital twin monitoring system, including: constructing a virtual mirror of the entire production line using a 3D modeling engine on a host computer interface, and rendering the flow status of materials across the entire line in real time; the system renders the movement trajectory of each tray in the virtual space based on the real-time calculated line velocity and predicts the time to reach the downstream process; the system marks the load status of each process on the 3D interface with different colors, where the first color represents normal synchronization of the entire line's cycle time, the second color represents that the predicted load of the downstream process is too high and has triggered the feed current limit, and the third color represents that it has entered a closed-loop handling state of adaptive speed reduction and power compensation; the digital twin monitoring system also integrates an augmented reality interface to overlay real-time cycle time data, current waveforms, and material position information on top of the physical equipment.

[0029] Preferably, the method includes a model predictive control strategy for long-cycle thermal inertia: the controller, based on a first-order transfer function model with pure time delay, predicts the furnace temperature field changes within a future preset time period according to the temperature fluctuation trend within a preset historical time period and the current roller speed adjustment; if the prediction result deviates from the process curve, the system pre-sets an intervention time to adjust the speed of the circulating fan or the set point of the heating group, using advanced control to offset the temperature overshoot or lag caused by thermal inertia; the system also receives meteorological data from the factory location, including atmospheric pressure, humidity and outdoor temperature, and uses environmental parameters as feedforward compensation to fine-tune the bias value of the heating power in advance, ensuring the robustness of the production line cycle under different climatic conditions.

[0030] Preferably, the method employs a high-precision positioning and calibration system in the material conveying stage: high-temperature resistant passive radio frequency identification (RFID) readers are installed at the feed inlet and discharge outlet of the roller kiln, and a unique electronic tag is embedded in the bottom of each container; the system records the entry time, exit time, and timestamp of each container passing through the junction of each heating section in real time; the timestamp data is cross-checked with the theoretical position calculated by the main control cycle; if the deviation between the physical position and the logically calculated position exceeds a preset deviation threshold, the system automatically triggers the cycle calibration logic to correct the speed mapping coefficient; when the downstream process encounters an anomaly and requires low-speed operation, the system feeds back the signal to the upstream sand milling process, and the sand mill control system automatically reduces the slurry circulation flow rate according to the buffer capacity of the spray drying tower.

[0031] Example 1: To make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments.

[0032] In the distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time, precise control of automated production lines with high thermal inertia and long process paths, such as lithium iron phosphate, is achieved by establishing highly synchronized physical entities and logical control associations. The specific implementation process of this embodiment is as follows:

[0033] Step 1 anchors the main control cycle of the entire distributed manufacturing line. At the core of the distributed manufacturing system, namely the roller kiln sintering process, the rotational frequency is collected in real time by an incremental photoelectric encoder installed on the shaft end of the main drive motor. The encoder resolution is set to 2048 lines, and the sampling period is fixed at 10 milliseconds. The controller receives the A / B phase differential pulse signal output by the encoder through the counting module, and extracts the real-time angular velocity of the motor rotor using a 4x frequency multiplication processing logic. This real-time rotational frequency is converted into an analog or digital quantity reflecting the physical rotational speed of the roller shaft by the signal conversion module of the field control unit, and is defined as the heartbeat signal for the entire line to coordinate.

[0034] Based on the physical structural parameters of the main drive roller, including the roller diameter, the transmission ratio of the reducer, and the correction coefficient of the chain drive system, the controller calculates the current operating linear speed. Specifically, the operating linear speed is obtained by multiplying the rotational frequency by the roller circumference and dividing by the total transmission ratio. To achieve quantitative mapping of production capacity, the system pre-configures the rated loading mass data of a single tray in its memory, for example, 5 kg of lithium iron phosphate precursor per tray. By monitoring the distribution density of trays per unit length at the feed end, for example, 4 trays per linear meter, the linear speed is multiplied by the loading mass and distribution density to obtain the real-time production capacity output per unit time.

[0035] In this process, real-time production output is measured in kilograms per hour. The calculated production data is not only used for real-time display but also serves as a global reference benchmark. The controller distributes this reference signal to the front-end sand mill slurry metering pump controller and the back-end screw feeder controller via an industrial network based on a real-time Ethernet protocol. Upon receiving the reference signal, each controller performs local closed-loop adjustment using a preset proportional coefficient to ensure that all physical units in the distributed system are strictly synchronized on the time axis.

[0036] Step 2 involves constructing a cross-process load prediction model. The system establishes a chain-like collaborative model encompassing the inventory of the sand milling process, the power of the spray drying process, the roller speed of the sintering process, and the frequency of the air-crushing process. This model achieves feedforward control by monitoring the material level status of the unloading hopper of the front-end spray drying tower. Three sets of 80GHz frequency-modulated continuous wave radar material level gauges are symmetrically installed on the top of the unloading hopper to measure the spatial distance on the material surface; simultaneously, four sets of pressure strain sensors are installed on the conical hopper wall of the unloading hopper to sense the gravity distribution of the material.

[0037] The controller employs a weighted average algorithm to fuse multiple radar and pressure signals, filtering out measurement errors caused by changes in the angle of repose of powder materials, material buildup, or dust. The model calculates the material level height and its rate of change in real time, i.e., the derivative of the material level with respect to time. When the material level in the unloading hopper exceeds the preset high-level warning threshold, the control logic determines that there is excess output at the front end. At this time, the system sends an acceleration command to the frequency converter at the roller kiln feed end via the fieldbus, increasing the discharging frequency by a preset 5% to 10% to accelerate the material's entry into the sintering stage. Conversely, when the material level is below the preset low-level threshold, the discharging frequency is simultaneously reduced to prevent the kiln from operating under no-load conditions.

[0038] Furthermore, the chain-based collaborative model incorporates feedforward correction logic for material moisture content. An online infrared moisture analyzer is installed at the discharge outlet of the spray drying tower, collecting material moisture data every second. If the detected powder moisture content exceeds the preset standard value of 1%, the model automatically triggers a correction mechanism, reducing the capacity output mapping coefficient. At this point, based on the moisture deviation value, the system extends the physical dehydration time and chemical reaction time of the material in the high-temperature zone of the kiln by reducing the roller speed.

[0039] Step 3 implements multi-source data collaboration and status monitoring. The system constructs a multi-source data interaction matrix covering the entire process through an industrial communication network based on the Profinet protocol. The rows of this matrix represent various physical process nodes such as sand milling, spraying, sintering, and gas crushing, while the columns represent sensor parameters such as pressure, temperature, current, frequency, flow rate, and vibration. The data synchronization cycle is controlled within 20 milliseconds.

[0040] The controller utilizes principal component analysis (PCA) to perform real-time dimensionality reduction on the high-dimensional interaction matrix. By performing eigenvalue decomposition on the covariance matrix, the top three key principal components contributing over 85% to system stability are extracted and defined as the system's eigenvectors. For the critical backend stage, namely the airflow pulverization process, the system focuses on collecting the three-phase current values ​​of the pulverizing disc drive motor, the opening feedback of the feed rotary valve, and the inlet and outlet pressure difference of the cyclone separator.

[0041] The system establishes a correlation function to describe the logical relationship between the current fluctuation rate of the air-jet motor and the bulk density of the material. When high-frequency oscillations in the motor current are detected and the average value shows an upward trend, while the differential pressure of the cyclone separator increases, the feature vector will shift and point to the abnormal load area. This feature extraction method based on multi-source data fusion enables minute-level advance detection of material blockage tendencies in downstream processes, rather than relying on post-event shutdown alarms.

[0042] Step 4 involves abnormal closed-loop handling and adaptive adjustment. When the current value of the downstream airflow pulverizing process exceeds 115% of the rated current and lasts for more than 3 seconds, the system determines it as a slight overload or material blockage warning. At this time, the control center activates the adaptive deceleration algorithm, which no longer executes the traditional emergency shutdown logic, but instead maintains the operation of each heating system of the roller kiln.

[0043] The controller issues a command to gradually reduce the main drive frequency of the roller kiln by 2Hz per minute until it reaches the preset safe frequency lower limit. Simultaneously, the system reduces the rotational speed of the feed feeder by adjusting the front-end feed inverter in reverse, and may even temporarily stop the feeding action, thereby reducing the subsequent material flow into the kiln at the source.

[0044] To address the thermal inertia issue caused by the rate of decrease, the system implements power compensation adjustment. As the linear velocity decreases, the residence time of the material in the furnace increases proportionally. If the power is not adjusted, the material will receive excessive thermal radiation, leading to overheating. Based on the current rate of decrease, the system invokes a preset heat balance equation to calculate the power reduction coefficient for each temperature zone in real time. For high-temperature zones above 1000 degrees Celsius, the system adjusts the thyristor duty cycle of the silicon carbide or silicon molybdenum rod heaters to synchronously reduce the heating power. This power compensation logic ensures that the total energy absorbed by the material remains consistent with the standard process curve even under variable-cycle operation.

[0045] The anomaly handling also includes self-recovery logic for current recovery rate. When the current of the downstream air jet mill drops below 90% of its rated value and the differential pressure of the cyclone separator returns to normal, the system determines that the blockage risk has been eliminated. At this time, the control system initiates a stepped speed-up scheme, divided into an initial speed recovery stage, a medium-speed stabilization stage, and a full-load recovery stage. A 180-second observation period is set after each speed-up stage. If the current fluctuation rate exceeds 5% during this period, the speed-up is immediately stopped.

[0046] This embodiment also integrates a digital twin monitoring system based on WebSockets communication technology. In the host computer system, a virtual mirror of the entire production line is constructed using a 3D modeling engine. Based on the real-time linear velocity calculated in step 1, the system renders the movement trajectory of each pallet in the virtual space. Operators can intuitively observe the material distribution density in each temperature zone inside the kiln. The system uses different colors on the 3D interface to indicate the process status: green represents normal synchronization of the entire line's cycle time, yellow represents a high predicted load at the back end that has triggered feed flow restriction, and red represents a closed-loop processing state of adaptive speed reduction and power compensation.

[0047] To address the long-cycle thermal inertia, this embodiment employs a model predictive control strategy. The controller incorporates a transfer function model based on a first-order element with pure time lag. By monitoring the temperature fluctuation trend over the past 10 minutes and the current frequency adjustment, it predicts the furnace center temperature change over the next 5 minutes. If the predicted value exceeds the process requirement of ±5 degrees Celsius, the system will adjust the speed of the circulating fan or the setpoint of the heating element in advance, using proactive control to counteract the lag effect of thermal inertia.

[0048] At the physical layer of material conveying, the system employs RFID high-precision positioning technology. High-temperature resistant passive RFID readers are installed at the feed inlet, the junctions of each heating section, and the discharge outlet of the roller kiln. Each container is embedded with a unique RFID electronic tag at its bottom. When a container passes the reader, the system records its precise timestamp and location information. These measured data are cross-checked in real time with the theoretical position calculated based on the master control cycle. If the deviation between the physical position and the logically calculated position exceeds 50 mm, the system determines that the drive chain is slipping or the roller is worn, automatically triggering the cycle calibration logic to correct the speed mapping coefficient.

[0049] Furthermore, this embodiment also achieves deep linkage with the front-end sand milling process. When the system is in speed reduction and risk avoidance mode, control commands are sent to the variable frequency drive of the sand mill through a reverse feedback link. The sand mill automatically reduces the outlet pressure of the slurry circulation pump and the speed of the main motor, and dynamically adjusts the slurry flow rate according to the buffer capacity of the spray drying tower. This closed-loop cycle coordination ensures that the entire production line is under uniform frequency constraints from the initial wet grinding to the final dry pulverization.

[0050] Example 2: In another preferred embodiment, the distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time further improves the system's anti-interference capability in complex electromagnetic environments by introducing a multi-level redundant control architecture.

[0051] For the anchoring of the main control cycle in step 1, this embodiment adds an extra set of radar tachometers as redundant sensors at the non-drive end of the roller kiln. This radar tachometer utilizes the Doppler effect, transmitting 24GHz microwaves and receiving the reflected signals from the roller surface to directly obtain the linear velocity of the roller. The controller integrates a signal optimization module that automatically selects the optimal signal source as the main control cycle by calculating the signal-to-noise ratio and consistency between the encoder signal and the radar signal in real time. If the deviation between the two sets of signals exceeds 3%, the system will trigger a sensor failure warning and switch to a safe operating mode based on historical average frequencies.

[0052] In the cross-process load prediction model of step 2, this embodiment adds a real-time calibration step for material density. An online nuclear scale or electronic screw scale is installed below the discharge hopper of the spray drying tower. This device measures the instantaneous flow rate of material entering the feeder in real time. The controller compares this mass flow rate with the volumetric flow rate calculated in step 1 and dynamically corrects the bulk density parameter of the material. This correction allows the prediction model to adapt to the differences in flowability and bulk density of different batches of precursor powder, thereby achieving more precise adjustment of the discharging frequency.

[0053] For the multi-source data collaboration in step 3, this embodiment employs a distributed database storage architecture. All sensor data is simultaneously uploaded to the central controller and stored on industrial-grade storage cards at the edge of each process step. The system uses a time-series database-based data compression algorithm to perform lossy compression on high-frequency fluctuation data such as frequency and current, while automatically switching to lossless recording mode when a data mutation is detected. The construction of the data interaction matrix is ​​no longer limited to a single linear correlation but introduces nonlinear correlation analysis, using mutual information indicators to measure the coupling strength between parameters of different processes.

[0054] In the abnormal closed-loop handling process of step 4, this embodiment refines the adaptive speed reduction execution strategy. The system divides the roller kiln into a preheating zone, a high-temperature reaction zone, and a cooling zone. When a risk of material blockage occurs at the downstream end and speed reduction is required, the power compensation logic of different temperature zones executes differentiated instructions. Since the preheating zone is less sensitive to the thermal history than the high-temperature reaction zone, the system prioritizes maintaining the temperature field stability of the high-temperature zone, while achieving overall energy consumption balance by increasing the power reduction in the preheating zone. At the same time, before issuing the speed reduction instruction, the system will send a frequency increase signal to the induced draft fan of the airflow mill in advance, thereby increasing the system's air volume to assist in clearing any potential material accumulation.

[0055] This embodiment also includes a fault diagnosis module based on edge computing. An edge computing unit with neural network inference capabilities is deployed within the control cabinet of each process. This unit analyzes the spectral characteristics of the drive motor in real time, extracting characteristic frequencies representing bearing wear, gear slippage, or load imbalance. When the amplitude of these characteristic frequencies exceeds a threshold, the system automatically adds minute oscillation compensation to the global cycle time, actively disrupting the bridging tendency of materials within the pipeline.

[0056] In terms of communication security, the multi-source data collaborative network employs an identity authentication mechanism based on hardware encryption chips. All cycle commands issued to the frequency converter and control components are digitally signed to prevent unauthorized speed increases or shutdowns due to network attacks. The system also features a mechanical safety protection chain: when the electronic control logic completely fails and the kiln linear speed exceeds the rated value by 20%, the mechanical speed limiter will forcibly cut off the main drive power and automatically activate the emergency cooling fan to protect the expensive ceramic rollers from deformation due to high temperatures.

[0057] Example 2 also specifically optimized the smooth transition logic of the cycle time during material switching. When the production line needs to switch from producing high-nickel ternary materials to lithium iron phosphate materials, the system automatically adjusts the main control reference frequency, feed density, and power distribution of each temperature zone by calling the preset formula library. At this time, the cycle time control system enters the "gradual switching" mode, calculates the position of the material interface in the kiln in real time, and updates the control parameters one temperature zone and one process as the material advances, realizing product switching without stopping the machine or emptying the kiln.

[0058] Example 3: In another preferred embodiment, the distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time achieves higher-dimensional intelligent management and control through deep integration of expert systems.

[0059] In this embodiment, particle morphology analysis based on machine vision is introduced into the load prediction model in step 2. A high-speed camera and a ring-shaped cold light source are installed at the discharge port of the spray drying tower to capture real-time images of the falling powder particles. The average particle size, roundness, and surface roughness are extracted using image processing algorithms. These morphological parameters are used as hidden variables input into the chain-like collaborative model. Experimental data shows that materials with lower particle roundness have higher frictional resistance during the subsequent spiral conveying and airflow pulverization processes. Therefore, when irregular particle morphology is detected, the system automatically lowers the preset current warning threshold in advance and appropriately reduces the feeding frequency to prevent potential blockage risks.

[0060] In the data collaboration section of step 3, the system establishes a global material tracking index. Each batch of material entering the system is assigned a unique virtual ID, which is associated with the slurry fineness during sand milling, the inlet and outlet temperatures during spraying, and the real-time roller speed and power during sintering. This multi-source data is archived to the cloud server in real time. Through big data analytics, the system can identify the underlying causes of abnormal gas-jet current in the downstream process. For example, if multiple material blockages are found to be related to a shortened sand milling time, the system will automatically correct the minimum running time constraint of the sand milling process, achieving cross-process quality backtracking and parameter self-optimization.

[0061] For the adaptive adjustment in step 4, this embodiment develops a parameter optimization algorithm based on reinforcement learning. The system performs simulation training based on historical operating data, learning how to find the optimal Pareto solution between the rate reduction and power compensation under different levels of current load. In this mode, when an anomaly occurs, the controller generates multiple adjustment schemes in real time and evaluates the predicted impact of each scheme on the final product's electrochemical performance. Ultimately, it executes the adjustment path that eliminates the risk of material blockage while maximizing production capacity.

[0062] In Example 3, the power compensation logic evolved into a full-field closed-loop control based on infrared thermal imaging. Inside the kiln, a multi-point infrared temperature measurement array was arranged along the length to scan the temperature distribution gradient of the rollers and the surface of the kiln carrier in real time. When the roller speed was reduced, the system no longer relied solely on theoretical equations to calculate the power, but instead directly performed independent PID fine-tuning on each group of heating elements based on the measured surface heat flux density distribution. This compensation method based on measured thermal field feedback eliminates the influence of external disturbances such as kiln insulation layer aging and ambient temperature fluctuations on control accuracy.

[0063] Furthermore, this embodiment also involves the coordinated optimization of energy consumption on the production line. Based on the synchronization of the entire line's cycle time, the system reduces the peak power load of the entire plant by adjusting the start-up sequence and operating phase of the high-power motors in each process. For example, when the roller kiln experiences a decrease in heating power due to abnormal speed reduction, the system automatically coordinates the front-end sand mill to perform high-load slurry storage, utilizing surplus transformer capacity to complete high-energy-consuming operations. This cycle-driven energy efficiency coordinated management significantly reduces the overall energy consumption per unit product while ensuring stable production.

[0064] In Example 3, the digital twin monitoring system integrates an augmented reality (AR) interface. When maintenance personnel enter the workshop wearing AR headsets, the system overlays real-time cycle data, current waveforms, and material location information directly onto the physical equipment. When a backend process triggers adaptive speed reduction logic, the AR system guides personnel to quickly locate the observation port where material accumulation may occur and displays the current process offset. This virtual-real interaction method greatly improves the anomaly response speed and maintenance efficiency in a distributed manufacturing environment.

[0065] In controlling long-cycle thermal inertia, Example 3 introduces weather sensing and prediction. The controller acquires real-time meteorological data of the factory location, including atmospheric pressure, humidity, and outdoor temperature, via an API interface. For non-sealed spray drying systems and roller kilns that require communication with the outside atmosphere for heat exchange, these environmental factors directly affect the heat loss rate. The system uses environmental parameters as feedforward compensation to fine-tune the bias value of the heating power in advance, ensuring the robustness of the production line cycle time under winter-summer transitions or rainy weather.

[0066] In Example 3, the closed-loop feedback mechanism was extended to the coating and automated packaging stages. When the overall production line slows down due to a backend anomaly, the system synchronously sends a signal to the finished product warehouse to adjust the operating cycle of the automatic palletizer and the cycle time of the vacuum packaging machine. This end-to-end frequency synchronization from source to end creates a flexible manufacturing ecosystem with self-regulating capabilities, enabling the entire lithium iron phosphate production line to dynamically adjust its metabolic frequency according to internal load and external demand, much like a single living organism.

[0067] Example 4: To further verify the reliability of the distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time, this example describes a distributed logical architecture implementation based on an edge gateway.

[0068] At the hardware level, this method deploys high-performance edge computing gateways at each major process node. These gateways exchange data at high speed with the PLCs of each process via an internal PCIe bus. The master control cycle time in step 1 is distributed to each edge gateway via a hard-wired synchronization signal, ensuring nanosecond-level clock synchronization accuracy. A real-time Linux operating system runs internally within the gateways, ensuring the deterministic execution of the load prediction algorithm and the adaptive speed reduction algorithm.

[0069] In step 2, the cross-process load prediction model employs a lightweight Long Short-Term Memory (LSTM) network for real-time inference in the edge gateway. By inputting historical time-series data such as material level, frequency, and power over the past 30 minutes, it predicts the material flow trend for the next 10 minutes. This deep learning-based prediction method can capture nonlinear characteristics in powder rheology, such as sudden collapses following the "mouse hole" effect in the silo wall, thereby allowing for advance adjustment of the feeding cycle and preventing damage to downstream processes caused by instantaneous material impacts.

[0070] In step 3, during data collaboration, the edge gateway uses the Message Queuing Telemetry Transport Protocol (MQTT) to upload feature vectors to the local server. The server constructs a distributed knowledge graph that maps equipment status, process parameters, and product quality indicators. When the differential pressure of the cyclone separator in the downstream air jet mill exhibits periodic small fluctuations, the expert system automatically diagnoses, through the knowledge graph, that this may be due to uneven particle size distribution of the upstream spray material. The system immediately feeds back to the atomizer frequency control module of the spray drying tower, optimizing particle size by fine-tuning the atomizing disc rotation speed, thereby addressing the load fluctuation problem at its root.

[0071] In step 4, the power compensation adjustment incorporates thermal radiation shielding compensation logic, as seen in Example 4. During the deceleration process of the roller kiln, it is necessary to consider not only the changes in energy absorbed by the material but also the impact of variations in the density of the carrier plates within the furnace on radiative heat transfer. The system pre-calculates the viewing angle coefficients for different carrier plate spacings using finite element analysis and performs secondary compensation for the set temperatures of each temperature zone through a correction coefficient table during actual adjustment. This meticulous thermodynamic consideration ensures that even during extremely low-speed operation for safety, the crystal growth process of the material remains within the optimal kinetic window.

[0072] For the recovery process after handling anomalies in a closed loop, this embodiment employs a smooth curve planning algorithm. The acceleration process follows an "S"-shaped curve, meaning that the acceleration is smaller at the start of acceleration and when approaching the target speed, while the acceleration is larger in the middle stage. This planning effectively avoids impact vibrations in the mechanical transmission system and also gives the thermal compensation system sufficient response time to adjust power according to speed changes. Throughout the self-recovery process, the system monitors and calculates the energy efficiency index of the entire line in real time. If an abnormal increase in energy consumption per unit output is detected, the system will automatically lock the current frequency for fault diagnosis.

[0073] Example 4 also focuses on the precise management of material residence time. The controller maintains a dynamic shift register to simulate the material's propulsion process within the kiln. Each bit of the register represents a specific length interval within the kiln. As the main control cycle ticks, material data moves bit by bit within the register. When the system slows down, the register automatically records the actual thermal history experienced by each segment of material. At the discharge end, the system classifies the material based on these records. For batches whose thermal history deviates significantly from the standard value due to excessively long slowdown times, the system automatically diverts them to the defective product bin via a pneumatic diversion valve, thus ensuring the absolute uniformity of the main production line's output.

[0074] In terms of software architecture, this embodiment adopts a microservice design. Functional modules such as load prediction, data matrix construction, speed-down algorithms, and power compensation run as independent containerized components. When an upgrade to the control logic of a certain process is required, it can be achieved seamlessly through blue-green deployment without interrupting the overall line cycle control. This modern software architecture design, combined with robust process cycle control logic, gives the method of this invention strong technological leadership and engineering scalability.

[0075] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of this invention is defined by the appended claims and their equivalents.

[0076] It will be understood by those skilled in the art that the above descriptions are merely preferred examples of the invention and are not intended to limit the invention. Although the invention has been described in detail with reference to the foregoing examples, those skilled in the art can still modify the technical solutions described in the foregoing examples or make equivalent substitutions for some of the technical features. All modifications and equivalent substitutions made within the spirit and principles of the invention should be included within the scope of protection of the invention. All technical features in this embodiment can be freely combined according to actual needs.

[0077] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time, characterized in that: Includes the following steps: Step 1, anchoring the main control cycle of the entire distributed manufacturing line: the operating frequency of the main drive roller is collected in real time by the sensor installed at the end of the main drive motor shaft of the roller kiln. The operating frequency is used as the heartbeat signal for the entire line to coordinate. Based on the structural parameters, transmission ratio and real-time frequency of the main drive roller, the current operating line speed is calculated and then mapped to the real-time production capacity output per unit time. The production capacity output is measured in a preset mass unit. The specific formula for calculating the linear velocity is as follows: ； in, This represents the corrected output linear velocity of the main drive roller. This represents the initial linear velocity set based on the baseline capacity. This represents the moisture correction factor. This represents the measured moisture content of the material. Represents the standard moisture content threshold. Represents the target production capacity constant. This represents the currently measured load per unit length. Step 2, construct a cross-process load prediction model: establish a chain collaborative model covering sand mill inventory, spray power, roller speed and air crushing frequency. By real-time monitoring of the material level height and material level change rate of the unloading bin of the front spray drying tower, predict the material flow demand of the subsequent process, and dynamically adjust the frequency converter operation rhythm of the roller kiln feed end according to the prediction results, so that the unloading frequency and the front feeding speed are logically matched. Step 3, Implement multi-source data collaboration and status monitoring: Real-time collection of operating parameters of sand milling, spraying, sintering and air crushing processes through industrial communication network, establish multi-source data interaction matrix, focus on monitoring the main motor current value, feed valve opening and cyclone separator pressure difference of the back-end airflow crushing process, and perform dimensionality reduction processing on the interaction matrix through principal component analysis algorithm to extract feature vectors representing load status; Step 4, Abnormal Closed-Loop Handling and Adaptive Adjustment: When the characteristic vector of the downstream airflow pulverizing process triggers the current overload or the material blockage warning threshold, the control system, while maintaining the continuous operation of the roller kiln, starts the adaptive deceleration algorithm to fine-tune the frequency of the main drive roller. At the same time, according to the deceleration ratio, the power compensation amount of each heating temperature zone inside the kiln is synchronously adjusted through the heat balance equation. Under the premise of ensuring the consistency of the sintering heat process of the material, the non-stop closed-loop collaborative control of the entire production line is realized. The specific power compensation calculation formula is as follows: ； in, This represents the adjusted heating power. This represents the rated heating power for the corresponding temperature zone. Represents the current actual linear velocity. Represents the rated linear velocity. Represents the heat exchange influencing factor. This represents the deviation in temperature field prediction. This represents the thermal response time constant of the kiln.

2. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, The process of anchoring the main control cycle of the entire line in step 1 includes: using an incremental photoelectric encoder with a predetermined resolution to monitor the roller rotation speed, and setting the sampling period to a preset sampling time interval; the controller receives the phase differential pulse signal output by the incremental photoelectric encoder through the counting module, and extracts the real-time angular velocity of the motor rotor using frequency multiplication processing logic; the controller performs frequency-voltage conversion processing on the collected pulse signal to obtain an analog quantity reflecting the physical linear velocity; the linear velocity is multiplied by the loading capacity of a single tray and the distribution density of trays per unit length to obtain the real-time material flow rate; the material flow rate is used as a global reference reference signal and distributed to the front-end metering pump controller and the back-end screw feeder controller through an industrial network based on the real-time Ethernet protocol. After receiving the reference signal, each controller performs local closed-loop adjustment through a preset proportional coefficient to ensure that each physical unit in the distributed system is synchronized on the time axis.

3. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, The process of establishing the cross-process load prediction model in step 2 includes: using multi-sensor fusion technology to monitor the material level in the unloading hopper of the spray drying tower; installing multiple sets of frequency-modulated continuous wave radar material level gauges on the top of the unloading hopper to measure the spatial distance on the material surface; and installing multiple sets of pressure strain sensors in the conical hopper of the unloading hopper to sense the gravity distribution of the material; using a weighted average algorithm to fuse multiple sets of radar signals and pressure signals to filter out measurement errors caused by uneven material accumulation, hanging material, or dust; the model calculates the material level height value and its derivative with respect to time in real time to obtain the material level change rate; when the material level in the unloading hopper is detected to exceed the preset high-level warning threshold, the system sends an acceleration command to the roller kiln feed frequency converter through the fieldbus to increase the discharging frequency by a preset adjustment range to accelerate material digestion; when the material level in the unloading hopper is detected to be lower than the preset low-level threshold, the discharging frequency is simultaneously reduced; The cross-process load prediction model uses the following formula to dynamically predict the material flow requirements of subsequent processes: ； in, Represents prediction of the future Material demand flow after time, This represents the material inflow rate corresponding to the current feeding cycle. This represents the rate of change in material level in the unloading hopper. This represents the current material level in the unloading hopper. This represents the target material level setting value. This represents the current operating power of the spray drying tower. , , These are weighting coefficients obtained through experience or identification; According to claim 1, a distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time is characterized in that the chain collaboration model involves a feedforward correction logic for the material moisture content, including: acquiring the dryness and humidity data of the sprayed material in real time by installing an online infrared moisture meter at the outlet of the spray drying tower; when the moisture content of the material represented by the dryness and humidity data is detected to exceed a preset moisture standard value, the chain collaboration model automatically corrects the capacity output mapping coefficient and extends the dehydration time of the material in the kiln by reducing the roller speed; the feedforward correction logic calculates the corrected output linear speed of the main drive roller based on the moisture deviation value, the baseline speed, the moisture correction coefficient, and the target capacity constant, realizing dynamic speed adjustment based on the material properties, and ensuring that the material received by the downstream airflow pulverization process is in a fragile and dry state.

4. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, In step 3, the industrial communication network involved in multi-source data collaboration adopts the real-time Ethernet protocol, and the data synchronization cycle is within a preset cycle range. The rows of the multi-source data interaction matrix represent the nodes of each physical process, and the columns represent the dimensions of pressure, temperature, current, frequency, flow rate, and vibration parameters. The system extracts the key principal components that contribute more than a preset proportion to the system stability by performing eigenvalue decomposition on the covariance matrix and defines them as the feature vector. For the back-end airflow pulverization process, the system collects the pulverizing disc speed, classifying wheel frequency, active power of the main drive motor, three-phase current value, and opening feedback of the feed rotary valve in real time. By establishing a correlation function between current fluctuation rate and material bulk density, the system realizes early detection of material blockage tendency. The following multi-source data fusion formula is used to quantify the load status of the back-end processes: ； in, This represents the overall load index of the back-end processes. This represents the current of the main motor for airflow pulverization. Represents the rated current. This represents the pressure difference between the inlet and outlet of the cyclone separator. Represents the maximum permissible differential pressure. This represents the vibration amplitude of the crushing disc. This represents the normal vibration reference value. , , These are the weighting coefficients, and their sum is 1.

5. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, The logic of the adaptive speed reduction algorithm in step 4 includes: when the main motor current of the air jet milling process reaches a preset current threshold and the duration exceeds a preset time threshold, the system determines that there is a slight overload risk; the controller gradually reduces the main drive frequency of the roller kiln according to a preset frequency gradient through a graded voltage reduction frequency adjustment strategy until it reaches the preset safe frequency lower limit; at the same time, the system reduces the rotation speed of the feed feeder and reduces the density of the load entering the kiln from the source by adjusting the front-end feed inverter in the opposite direction; the abnormal closed-loop handling also includes self-recovery logic based on the current recovery rate. When the main motor current of the back-end air jet milling process falls below the preset safe value and the fluctuation rate tends to stabilize, the system starts a step-by-step speed increase scheme. The graded voltage reduction frequency adjustment strategy uses the following dynamic step size formula to determine the rate reduction amount at each step: ； in, This represents the frequency change of the next deceleration step. Represents the baseline deceleration step size. This represents the current current of the main motor for airflow pulverization. Represents the overload warning threshold current. Represents the rate of change of current. , A positive adjustment coefficient; Furthermore, the stepped acceleration scheme uses the following smooth curve formula to set the target frequency during the recovery process: ； in, Represents the target frequency at the current moment. This represents the lower safety limit frequency during anomaly handling. Represents the target recovery frequency. This represents the acceleration time constant.

6. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, The power compensation adjustment and speed reduction algorithm are strongly coupled control logic, including: the system calculates the power reduction coefficient corresponding to each temperature zone in real time according to the preset ratio of the current speed reduction through the preset heat balance equation; when the roller speed decreases by a preset amount, the set temperature of the high temperature zone is slightly adjusted downward by the preset temperature deviation, or the heating power is reduced synchronously by adjusting the duty cycle of the thyristor of the heating control component; the heat balance equation is based on the current actual linear speed, rated linear speed, heat exchange influence factor, temperature field prediction deviation and the thermal response time constant of the kiln to perform power compensation calculation, ensuring that the total energy absorbed by the material under the variable cycle operation state is consistent with the standard process curve, and eliminating the temperature field deviation caused by speed change; A power compensation correction based on thermal history consistency is introduced, specifically: the cumulative heat absorbed by the material in the kiln is calculated in real time, and the heating power is additionally corrected according to the deviation from the standard heat. The correction formula is as follows: ； ； in, This represents the cumulative heat absorbed by the material inside the kiln. , These represent the times when materials enter and leave the kiln, respectively. Represents real-time heat exchange efficiency. This represents the target heat that the material should absorb under standard processes. This represents the power compensation correction amount. Represents the thermal inertia compensation coefficient. This represents the rated power for the current temperature range.

7. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, The method integrates a digital twin monitoring system, including: constructing a virtual mirror of the entire production line using a 3D modeling engine on a host computer interface, and rendering the flow status of materials across the entire line in real time; the system renders the movement trajectory of each tray in the virtual space based on the real-time calculated line velocity and predicts the time to reach the downstream process; the system marks the load status of each process on the 3D interface with different colors, where the first color represents normal cycle time synchronization across the entire line, the second color represents a predicted high load at the downstream end that has triggered feed current limiting, and the third color represents a closed-loop handling state of adaptive speed reduction and power compensation; the digital twin monitoring system also integrates an augmented reality interface to overlay real-time cycle time data, current waveforms, and material position information onto the physical equipment.

8. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, The method includes a model predictive control strategy for long-cycle thermal inertia: the controller is based on a transfer function model with a first-order element and a pure time lag, and predicts the furnace temperature field changes in a future preset time period based on the temperature fluctuation trend and the current roller speed adjustment within a preset historical time period; if the prediction result deviates from the process curve, the system adjusts the speed of the circulating fan or the set point of the heating group in advance by preset intervention time, and uses advanced control to offset the temperature overshoot or lag caused by thermal inertia; the system also receives meteorological data from the factory location, including atmospheric pressure, humidity and outdoor temperature, and uses environmental parameters as feedforward compensation to fine-tune the bias value of the heating power in advance, ensuring the robustness of the production line cycle under different climatic conditions.

9. The distributed manufacturing multi-source data collaboration and closed-loop control method driven by process cycle time as described in claim 1, characterized in that, The method employs a high-precision positioning and calibration system in the material conveying process: high-temperature resistant passive radio frequency identification (RFID) readers are installed at the feed and discharge ports of the roller kiln, and a unique electronic tag is embedded in the bottom of each container; the system records the entry time, exit time, and timestamp of each container passing through the junction of each heating section in real time; the timestamp data is cross-checked with the theoretical position calculated by the main control cycle; if the deviation between the physical position and the logically calculated position exceeds a preset deviation threshold, the system automatically triggers the cycle calibration logic to correct the speed mapping coefficient; when the downstream process encounters an anomaly and requires low-speed operation, the system feeds back the signal to the upstream sand milling process, and the sand mill control system automatically reduces the slurry circulation flow rate according to the buffer capacity of the spray drying tower.

Images