Demand driven flexible production control system

Abstract

The present application is a demand-driven flexible production control system, and relates to the technical field of industrial automation production control, comprising: a demand signal analysis and mapping module; an energy dissipation rate real-time calculation module; a dissipation rate derivative constraint evaluation module; a reverse equivalent pressure transmission module; a decentralized beat adaptive module.In the present application, the macro demand signal of the upper information system is converted into a smooth and continuous production beat instruction, so that the production system can quickly respond to the dynamic changes of market demand, while avoiding mechanical impact on the actuator caused by sudden changes in the instruction; through real-time monitoring and fine analysis of the energy dissipation characteristics of the actuator, the system can generate dynamic constraint thresholds based on the actual operating conditions of the equipment, identify abnormal trends in the operating state of the equipment in advance, and take appropriate adjustment measures before obvious signs of equipment failure appear.

Description

Technical Field

[0001] This invention relates to the field of industrial automation production control technology, and in particular to a demand-driven flexible production control system. Background Technology

[0002] Traditional production control systems have evolved over many years through technological advancements and engineering practices, resulting in a relatively mature centralized scheduling and fixed-cycle control system. This system can stably and efficiently meet the production needs of large-volume standardized products, laying a solid foundation for the large-scale development of modern manufacturing.

[0003] However, in the face of dynamically changing market demands, how to achieve rapid global response and smooth local transition of production cycle time, while taking into account both the completion of production tasks and the long-term operational stability of the system, has become an important direction that needs to be further explored in the field of industrial automation.

[0004] Real-time monitoring and health management of equipment operating status are core supporting technologies for ensuring the continuous and reliable operation of production systems. Existing technologies widely employ various sensing methods such as vibration, temperature, current, and torque to collect equipment operating parameters and can achieve alarms and post-failure diagnosis for common equipment faults, effectively reducing production losses caused by sudden equipment failures. However, most current condition monitoring systems and production control systems are architecturally independent, and equipment status information is not integrated into the dynamic adjustment process of production cycle time in a real-time and in-depth manner. Furthermore, existing energy consumption monitoring technologies focus more on the statistics of total energy consumption and long-term trend analysis, lacking detailed research on the dynamic energy dissipation characteristics of equipment and their instantaneous rate of change, making it difficult to identify early equipment anomalies and potential operational risks in advance.

[0005] Distributed production control architectures are increasingly widely used in modern discrete manufacturing and process manufacturing systems due to their high flexibility, scalability, and fault tolerance. Existing distributed control systems endow each production node with a certain degree of autonomous decision-making and execution capabilities, effectively improving the system's response speed to local disturbances.

[0006] However, existing inter-node coordination mechanisms are mostly based on the exchange of production information such as work-in-process inventory levels, production completion status, and process progress, lacking a quantitative pressure transmission and global collaborative adjustment mechanism based on real-time equipment operational risks. When a production node experiences operational degradation, upstream nodes struggle to adjust their production rhythm in a timely manner, easily leading to work-in-process backlogs or increased production bottlenecks, affecting the overall operational efficiency and resilience of the entire production system. Summary of the Invention

[0007] The purpose of this invention is to propose a demand-driven flexible production control system to solve the above-mentioned problems.

[0008] To achieve the above objectives, the present invention adopts the following technical solution: Demand-driven flexible production control systems include: The demand signal parsing and mapping module is used to receive macro demand signals from the upper-level information system, convert the macro demand signals into initial target beat instructions, and distribute the initial target beat instructions to each decentralized node after smoothing them. The real-time energy dissipation rate calculation module is used to synchronously collect the input active power and output mechanical power of the actuator, calculate the dynamic energy dissipation rate by combining it with the pre-calibrated static inherent loss, and perform filtering processing on the dynamic energy dissipation rate. The dissipation rate derivative constraint evaluation module is used to receive the dynamic energy dissipation rate after filtering and calculate the first-order time change rate. It combines the actual mechanical angular velocity and real-time temperature to generate a dynamic constraint threshold. The first-order time change rate is compared with the dynamic constraint threshold to calculate the risk index. The reverse equivalent pressure transmission module is used to receive the risk index and convert it into an initial reverse equivalent pressure through nonlinear mapping, and then transmit the actual reverse equivalent pressure to the upstream process. It also receives multi-source pressure signals transmitted from the downstream process and fuses them to generate a comprehensive reverse equivalent pressure. The decentralized beat adaptive module receives the initial target beat command and the comprehensive reverse equivalent pressure to calculate the speed drop value. It subtracts the speed drop value from the initial target beat command to generate a dynamic synthetic beat command. After boundary clamping the dynamic synthetic beat command, it drives the physical actuator.

[0009] Preferably, in the demand signal parsing and mapping module, the specific steps for converting the macroscopic demand signal into the initial target beat command are as follows: After receiving macro-level demand signals from the upper-level information system, the integrity of the macro-level demand signals is verified by using a cyclic redundancy check algorithm, and timestamps are extracted to remove out-of-order data packets. Extract the remaining target output and remaining delivery time window from the macro demand signal, divide the remaining target output by the remaining delivery time window and combine it with the dynamically updated comprehensive efficiency compensation coefficient to calculate the global theoretical target cycle time; The initial target physical cycle time is obtained by multiplying the global theoretical target cycle time by the mechanical transmission conversion coefficient of the corresponding process. The initial target physical beat is smoothed by using a command shaping filter based on a nonlinear S-shaped acceleration and deceleration curve. During the acceleration increase phase, the acceleration is increased with a constant jerk; during the uniform acceleration phase, the maximum acceleration is maintained; and during the acceleration decrease phase, the acceleration is decreased with a constant negative jerk, thereby generating continuous initial target beat commands.

[0010] Preferably, in the real-time energy dissipation rate calculation module, the input active power and output mechanical power of the actuator are collected synchronously, and the dynamic energy dissipation rate is calculated by combining the pre-calibrated static inherent losses. The specific contents include: The instantaneous voltage and current of the three phases are obtained by voltage transformers and Hall current sensors. The instantaneous voltage and current of the three phases are multiplied and accumulated to obtain the instantaneous total power. The input active power is calculated by integrating and averaging the instantaneous total power within the sliding data window. The rotor position is obtained by using an optical encoder and the actual mechanical angular velocity is calculated differentially. Combined with the actual mechanical torque obtained by the torque sensor, the output mechanical power is calculated by multiplying the actual mechanical angular velocity and the actual mechanical torque. The formula for calculating the dynamic energy dissipation rate is as follows: ; in, The dynamic energy dissipation rate, For input active power, To output mechanical power, Current speed and temperature The static inherent loss.

[0011] Preferably, the process of filtering the dynamic energy dissipation rate is as follows: A statistical algorithm based on Kalman filtering is adopted. In the prediction step, the predicted value of the dissipation rate at the current time is calculated based on the final estimated value of the dissipation rate at the previous time step and the preset state transition law, and the prediction error covariance is calculated simultaneously. In the update step, the dynamic energy dissipation rate with noise calculated at the current moment is read as the observation value. The Kalman gain coefficient is dynamically calculated by combining the prediction error covariance and the measurement noise covariance. The Kalman gain coefficient is used to weight and fuse the predicted and observed values ​​of the dissipation rate to output a smooth final estimate. Trend analysis is performed on the final estimate, and the mean, variance, and peak characteristics within the preset time window are extracted and stored in a high-speed dual-port random access memory.

[0012] Preferably, the dissipation rate derivative constraint evaluation module calculates the first-order time rate of change in the following way: A fixed-length first-in-first-out data queue is constructed in memory as a sliding time window. Whenever the filtered dynamic energy dissipation rate is received, it is pushed to the head of the queue and the oldest data at the tail of the queue is pushed out. Using time as the horizontal axis and dynamic energy dissipation rate as the vertical axis, the coefficients of the constant term, the first term, and the second term are continuously adjusted using the least squares method to find a smooth fitting parabola that passes through all data points within the sliding time window. The slope of the tangent line of the smooth fitting parabola at the latest moment is extracted as the first-order time rate of change to suppress the amplification effect of high-frequency noise.

[0013] Preferably, the process of generating a dynamic constraint threshold by combining the collected actual mechanical angular velocity and real-time temperature, and comparing the first-order time rate of change with the dynamic constraint threshold to calculate the risk index is as follows: ; in, For the current moment Dynamic constraint threshold, The baseline dissipation rate derivative tolerance limit, This represents the current actual mechanical angular velocity. The highest speed designed for physical purposes. For real-time temperature, For standard operating temperature, The highest permissible temperature limit, This is the speed penalty coefficient. This is the temperature penalty coefficient; When calculating the risk index, the lower limit threshold for early warning is set as a fixed proportion of the dynamic constraint threshold. When the first-order time rate of change is less than or equal to the lower limit threshold for early warning, the risk index is assigned a value of zero. When the first-order time rate of change is greater than the lower warning threshold and less than the dynamic constraint threshold, the portion of the first-order time rate of change that exceeds the lower warning threshold is calculated and divided by the difference between the dynamic constraint threshold and the lower warning threshold to obtain the linear approximation ratio. The linear approximation ratio is then nonlinearly amplified to generate a risk index. When the first-order time rate of change is greater than or equal to the dynamic constraint threshold, the risk index is forcibly clamped to the maximum value of one.

[0014] Preferably, the reverse equivalent pressure transmission module converts the risk index into an initial reverse equivalent pressure and generates an actual reverse equivalent pressure, specifically including: The formula for calculating the initial reverse equivalent pressure is: ; in, For the initial reverse equivalent pressure, As a risk index, Based on the basic pressure gain coefficient, The coefficient of pressure expansion; The virtual connection between two adjacent processes is regarded as a second-order mechanical vibration system containing a mass block, a spring, and a damper. The initial reverse equivalent pressure is taken as the external excitation force acting on the virtual mass block. The virtual mass coefficient determines the inertia of the mass block to filter out high-frequency fluctuations, the virtual damping coefficient absorbs oscillation energy, and the virtual spring stiffness coefficient determines the magnitude of the steady-state transmitted pressure. The displacement of the virtual mass block is calculated by a numerical integration algorithm, and the displacement is multiplied by the virtual spring stiffness coefficient to generate a smooth actual reverse equivalent pressure.

[0015] Preferably, the specific steps for generating the integrated reverse equivalent pressure through fusion include: Extract the sending timestamp from the received multi-source pressure signal data packets transmitted by downstream processes, compare the sending timestamp with the local current time to calculate the actual transmission delay time, extract the pressure data points in the historical buffer to calculate the rate of change of the pressure signal at the time of transmission, add the received historical pressure value to the product of the rate of change and the actual transmission delay time, and deduce the real pressure state at the current time to complete the time alignment reconstruction. Based on the importance of downstream processes and physical buffer capacity, weight coefficients are assigned to the reconstructed multi-source pressure signals and weighted multiplication is performed. The maximum value envelope extraction logic is executed to compare and select the maximum value among the weighted pressure values ​​as the comprehensive reverse equivalent pressure.

[0016] Preferably, the decentralized beat adaptive module calculates the speed drop value by including: The formula used is as follows: ; in, The speed drop value, To comprehensively address the reverse equivalent pressure, This is a virtual compliance coefficient. This is the virtual damping coefficient; Within each control cycle, the received initial target beat command is read, and the calculated speed drop value is subtracted from the initial target beat command to obtain the preliminary correction beat command, ensuring that the actual target beat of the current process is lower than the macro demand beat when there is risk pressure downstream.

[0017] Preferably, when generating the dynamic synthesized beat command, if the speed drop value is detected to be in a decreasing phase, an exponential smoothing recovery mechanism based on a first-order inertial filtering algorithm is activated. The difference between the initial target beat command at the current moment and the final output beat of the previous cycle is calculated. The smoothing coefficient is calculated by dynamically calling the recovery time constant based on the physical moment of inertia and mechanical stiffness. The difference is multiplied by the smoothing coefficient to obtain the beat recovery increment and accumulated to the final output beat of the previous cycle. Boundary clamping of dynamically synthesized beat instructions includes: In amplitude clamping, if the dynamic synthesized beat command is greater than the maximum allowed physical beat, it is forcibly truncated and assigned the maximum allowed physical beat; if it is less than the minimum allowed physical beat, it is forcibly boosted and assigned the minimum allowed physical beat. In rate-of-change clamping, the actual required acceleration is obtained by dividing the instruction change in a single cycle by the control cycle. If the actual required acceleration exceeds the maximum allowable physical acceleration range, the instruction that drives the physical actuator is generated based on the actual output instruction of the previous cycle and the maximum allowable physical acceleration.

[0018] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention transforms macroscopic demand signals from the upper-level information system into smooth and continuous production cycle instructions, enabling the production system to quickly respond to dynamic changes in market demand while avoiding mechanical shocks to the actuators caused by sudden instruction changes. Through real-time monitoring and refined analysis of the energy dissipation characteristics of the actuators, the system can generate dynamic constraint thresholds based on the actual operating conditions of the equipment, identify abnormal trends in the equipment's operating status in advance, and take corresponding adjustment measures before obvious signs of equipment failure appear.

[0019] 2. This invention quantifies equipment operational risks into reverse equivalent pressure signals and achieves smooth transmission of these pressure signals and coordinated adjustment of upstream and downstream processes through a virtual second-order mechanical system model. When a production node experiences operational degradation, the risk pressure can be promptly transmitted to upstream processes, guiding upstream nodes to proactively adjust their production rhythm and avoid work-in-process backlog and the expansion of production bottlenecks. Employing a decentralized cycle time adaptive control architecture, each production node can make autonomous decisions based on local demand instructions and received comprehensive pressure signals, without relying on unified scheduling by a central controller, thus improving system response speed and fault tolerance. Simultaneously, through smooth recovery mechanisms and boundary clamping processing, a smooth transition of production cycles is ensured, optimizing the system's energy utilization efficiency. Attached Figure Description

[0020] Further details, features, and advantages of this application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 This is a system structure diagram of the present invention. Detailed Implementation

[0021] Several embodiments of this application will now be described in more detail with reference to the accompanying drawings to enable those skilled in the art to implement this application. This application may be embodied in many different forms and for various purposes and should not be limited to the embodiments set forth herein. These embodiments are provided to make this application thorough and complete, and to fully convey the scope of this application to those skilled in the art. The embodiments described do not limit this application.

[0022] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having a meaning consistent with their meaning in the relevant field and / or the context of this specification, and shall not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0023] Example 1

[0024] Its specific implementation method is combined with the appendix Figure 1 Please provide a detailed explanation.

[0025] Appendix Figure 1 The block diagram of the demand-driven flexible production control system provided in the embodiment of the present invention shows the connection relationship between the demand signal parsing and mapping module and the decentralized tachometer adaptive module, and marks the main functional interaction flow of each module.

[0026] In this embodiment, it includes: The demand signal parsing and mapping module is used to receive macro demand signals from the upper-level information system, convert the macro demand signals into initial target beat instructions, and distribute the initial target beat instructions to each decentralized node after smoothing them. The demand signal parsing and mapping module is the input terminal and macro-instruction conversion hub of the entire decentralized beat regulation system.

[0027] In traditional industrial control architectures, production demands issued by Enterprise Resource Planning (ERP) or Manufacturing Execution Systems (MES) are typically represented by discrete macro-level data such as order quantities, delivery deadlines, and product specifications. This data remains at the information management level and cannot be directly recognized and executed by the underlying Programmable Logic Controllers (PLCs) or motor drives.

[0028] The core task of this module is to receive these discrete macroscopic demand signals from the upper-level information system and, through a series of logical mapping and signal conditioning mechanisms, transform them into continuous physical quantity setpoints (i.e., initial target production cycle time, such as the linear speed of the conveyor belt, the rotational angular velocity of the spindle, the flow rate setpoint of the pump, etc.) that can be directly processed by the decentralized control nodes at the bottom level.

[0029] This module ensures that macroeconomic demand fluctuations can be smoothly and accurately mapped to the initial driving force of the underlying physical devices.

[0030] This module first establishes a highly reliable data communication link with the upper-level MES system via an industrial Ethernet bus (such as PROFINET, EtherCAT, or Modbus TCP protocol).

[0031] The module is equipped with a dedicated communication interface unit, which includes an independent network controller and physical layer transceiver, used to periodically poll or receive macroscopic demand data packets triggered by events.

[0032] The received raw data packet contains the target total output, the current output completed, the remaining delivery time window, and the product process formula number.

[0033] To prevent data packet loss or bit errors caused by strong electromagnetic interference in industrial environments from interfering with the underlying control, the module is equipped with a data verification and redundancy removal mechanism.

[0034] Specifically, the cyclic redundancy check algorithm is used to verify the integrity of the received data frames; The verification process involves dividing the data polynomial by a preset generator polynomial at the sending end and appending the remainder to the end of the data frame. After receiving the data, the receiving end performs the same division operation using the same generator polynomial. If the remainder is zero, it is determined that the data has not been flipped or lost during transmission.

[0035] Once a verification error is detected, the module will discard the frame data directly, retain the requirement instructions from the previous valid cycle, and send a retransmission request to the upper-layer system.

[0036] For continuously received valid data, the module extracts the timestamp from the packet header for comparison, and eliminates out-of-order packets caused by network routing delays, ensuring that the demand signal is strictly monotonically increasing in the time series, thereby providing a stable and reliable data source for subsequent beat mapping.

[0037] After acquiring the preprocessed and valid macroeconomic demand data, the module enters the core logical mapping stage. The purpose of this stage is to calculate the theoretical production cycle time to meet the current delivery and output requirements.

[0038] Taking into account the unavoidable minor equipment downtime, material changeover time, and yield fluctuations during the production process, the module introduces a dynamic comprehensive efficiency compensation coefficient.

[0039] This coefficient is a dimensionless parameter between 0 and 1, and its value is updated in real time by the system based on historical production data (such as the ratio of actual output to theoretical output in the past 24 hours).

[0040] This module uses the following logical steps to calculate the global theoretical target cycle time: The module extracts the current remaining delivery time window and converts it into a base time reference in seconds. The module extracts the total number of target outputs that have not yet been completed. Then, it divides the remaining target output by the remaining delivery time window to obtain a base production rate under ideal conditions. To compensate for efficiency losses in actual production, the module divides this base production rate by a dynamically updated comprehensive efficiency compensation coefficient, thus obtaining a global theoretical target cycle time amplified by losses. This cycle time represents the overall material flow rate that the entire production line must achieve at the current moment.

[0041] In a decentralized control architecture, a production line consists of multiple independent processes connected in series or parallel. Due to the different physical processing characteristics of each process (for example, the heating process has a long response time, while the stamping process has a short response time), the global theoretical cycle time must be further mapped to the initial target physical quantities of each specific process.

[0042] For each specific decentralized process, the module pre-sets a dedicated mechanical transmission conversion coefficient. This coefficient comprehensively considers physical and mechanical parameters such as the speed ratio of the motor reducer, the diameter of the transmission pulley, and the lead of the lead screw. The module directly multiplies the calculated global theoretical target cycle time by the mechanical transmission conversion coefficient of the corresponding process, thereby accurately converting the macroscopic "number of parts processed per second" into the "rotational radians per second" of the bottom-level motor spindle or the "millimeters per second movement" of the conveyor belt.

[0043] Through this logical mapping process, the macro-level order demand is precisely transformed into physical quantity settings that can be understood by each underlying execution mechanism.

[0044] In real-world production environments, the demand commands issued by the MES system often change abruptly. For example, when an urgent order arrives, the target output can surge instantly, causing a dramatic abrupt change in the calculated initial target physical cycle time.

[0045] If such a step signal is sent directly to the underlying actuator, it will cause the motor to output a huge transient electromagnetic torque, resulting in serious mechanical shock, gear breakage, conveyor belt slippage, or even material breakage and other physical damage.

[0046] To eliminate this step shock, this module must smooth and make the initial target physical beat continuous before issuing the command.

[0047] The module internally constructs a command shaping filter based on a nonlinear S-shaped acceleration / deceleration curve.

[0048] When a change in the target beat setting is detected, the instruction shaping filter does not immediately output a new setting value. Instead, it plans a smooth transition trajectory based on the preset maximum allowable acceleration and maximum allowable jerk (i.e., the rate of change of acceleration).

[0049] The trajectory is generated following the logic of physical kinematic constraints: In the first stage (acceleration increment stage), the acceleration of the control command starts from zero and gradually increases at a constant accelerometer until the maximum acceleration allowed by the equipment is reached. This stage ensures the smoothness of the equipment startup and avoids instantaneous impacts from mechanical gaps. In the second stage (uniform acceleration stage), the control command continuously increases the speed at a constant maximum acceleration to approach the target beat at the fastest safe rate. In the third stage (acceleration reduction stage), when the speed is about to reach the target set value, the acceleration gradually decreases with a constant negative jerk until the acceleration smoothly drops to zero. At this time, the speed stabilizes at the new target set value without overshoot.

[0050] Through this nonlinear S-curve shaping, the originally discrete and drastically abrupt macroscopic demand signals are transformed into continuous, smooth control commands that conform to the physical and dynamic limits of the underlying equipment.

[0051] The smoothed instructions need to be distributed to the various decentralized local controllers. This module uses a publish-subscribe communication model for instruction distribution.

[0052] The module acts as the publisher, broadcasting data frames containing timestamps, process numbers, and target cycle instructions to the control bus; Each local controller acts as a subscriber, extracting the corresponding control instructions based on its own process number.

[0053] In a decentralized architecture, each local controller operates independently. Without a unified time base, phase differences can arise in the response of different processes to changes in cycle time, leading to material accumulation or flow interruptions between processes. Therefore, this module also undertakes the responsibility of network-wide time synchronization. The module integrates a high-precision hardware clock source and adopts the IEEE 1588 precise time protocol.

[0054] The specific execution logic for time synchronization is as follows: The master clock of this module periodically sends synchronization messages with precise transmission timestamps to all local controllers; When the local controller receives the message, it records its local receiving timestamp; the local controller sends a delay request message to the master clock and records the sending timestamp. After receiving the request, the master clock records the received timestamp and sends the timestamp back to the local controller via a delayed response message.

[0055] The local controller uses these four timestamps to calculate the time difference between the round trip of the message in the network, thereby estimating the one-way delay time of the network transmission, and based on this, calculates the phase deviation between its own local clock and the master clock.

[0056] The local controller adjusts its own clock oscillator frequency based on this deviation to achieve alignment with the master clock.

[0057] This time synchronization mechanism ensures that all decentralized nodes can coordinate their actions within a unified time coordinate system when they receive the smoothed beat instructions.

[0058] The real-time energy dissipation rate calculation module is connected to the demand signal analysis and mapping module. It is used to synchronously collect the input active power and output mechanical power of the actuator, calculate the dynamic energy dissipation rate by combining the pre-calibrated static inherent loss, and perform filtering processing on the dynamic energy dissipation rate. The real-time energy dissipation rate calculation module is the core detection unit of this system for realizing underlying physical state perception and adaptive protection.

[0059] In flexible production, when the production cycle changes according to demand instructions (especially with a significant increase in speed), nonlinear physical factors such as mechanical friction, fluid resistance, and material deformation resistance inside the equipment will increase significantly.

[0060] Traditional control systems only focus on whether the motor speed or position follows the set value, ignoring the extra energy cost that the equipment expends to maintain that speed.

[0061] This kind of blind following often leads to equipment operating in an overloaded or sub-optimal state, accelerating mechanical fatigue and even causing sudden failures.

[0062] This module synchronously collects the input electrical power and output mechanical power of each process actuator at high frequency, and calculates the system's energy dissipation rate in real time.

[0063] This dissipation rate not only reflects the inherent losses of the equipment, but more importantly, it can extremely sensitively capture the sudden changes in additional physical resistance caused by changes in the beat rate, providing the most basic physical limit feedback signal for subsequent adaptive beat rate adjustment.

[0064] To accurately assess energy dissipation, it is first necessary to precisely measure the input electrical power of the actuator (such as a three-phase AC asynchronous motor or a permanent magnet synchronous motor).

[0065] This module deploys high-precision voltage transformers and Hall current sensors at the front end or inside the motor drivers of each process.

[0066] Hall effect sensors utilize the physical effect of magnetic fields deflecting electrons to measure large currents without contact and have extremely high response speeds.

[0067] Considering the large amount of electromagnetic interference in industrial environments and the high-order harmonics of pulse width modulation output from frequency converters, this module performs hardware conditioning on the acquired analog voltage and current signals.

[0068] The signal first passes through a passive RC low-pass filter with a cutoff frequency of 10kHz to filter out high-frequency switching noise, then enters the signal amplification and isolation circuit, and finally is digitally sampled by an analog-to-digital converter with a resolution of no less than 16 bits.

[0069] The sampling frequency was set to 20kHz to ensure that the transient waveforms of current and voltage could be fully captured.

[0070] After acquiring the digitized three-phase instantaneous voltage and current signals, the digital signal processor inside the module no longer relies on simple formula stacking, but instead executes the following power calculation logic: Within each microsecond-level sampling period, the processor multiplies the instantaneous voltage and instantaneous current of phase A, phase B, and phase C, and then accumulates these three products in real time to obtain the total instantaneous input power of the system at the current extremely short instant.

[0071] Due to the characteristics of alternating current, this instantaneous power contains a large amount of pulsating components and cannot be directly used for condition assessment. Therefore, the processor allocates a sliding data window in memory, the length of which strictly corresponds to one complete electrical cycle of the motor. The processor accumulates all the instantaneous power values ​​recorded in this window and divides them by the total number of data points in the window, thereby calculating a smooth input active power that has been filtered out of AC pulsations.

[0072] The calculation process is executed cyclically within the digital signal processor with extremely high real-time performance, ensuring continuous updates of the input electrical power data.

[0073] While acquiring the input electrical power, the module must simultaneously acquire the effective mechanical power output from the actuator to the load.

[0074] The measurement of mechanical power relies on physical sensors installed on the output shaft of the motor or the output end of the reducer.

[0075] This module uses a high-resolution photoelectric encoder or rotary transformer to measure the rotor position.

[0076] The module calculates the actual mechanical angular velocity by recording the change in the number of encoder pulses between two adjacent sampling periods and combining this with the sampling time interval.

[0077] To eliminate high-frequency noise caused by pulse counting discretization, the angular velocity signal is smoothed by a digital low-pass filter.

[0078] For measuring output mechanical torque, this module provides two parallel schemes to adapt to different industrial field conditions; Option 1: Install a strain gauge torque sensor directly on the drive shaft. When the drive shaft is subjected to torque and undergoes a slight deformation, the resistance value of the strain gauge changes. The resistance change is converted into a weak voltage signal through a bridge circuit. After amplification and calibration, the actual output torque is obtained.

[0079] Option 2: In compact devices where physical torque sensors cannot be installed, a full-dimensional state observer based on a mathematical model of the motor is built inside the module.

[0080] The observer uses known stator voltage, current, and rotor speed as inputs to simulate a virtual motor operating state within the software.

[0081] By continuously comparing the deviation between the output current of the virtual motor and the actual measured current, the observer uses a feedback gain matrix to correct the internal state variables of the virtual motor in real time, forcing the deviation to tend towards zero. When the deviation is eliminated, the electromagnetic torque estimated internally by the observer is considered equal to the actual electromagnetic torque.

[0082] The module subtracts the inertial torque required for the rotor's own acceleration from the electromagnetic torque, thereby indirectly and accurately observing the actual mechanical torque output to the load.

[0083] After obtaining the smooth mechanical angular velocity and the actual mechanical torque, the module multiplies the two in real time to calculate the effective mechanical power output by the actuator.

[0084] To ensure the accuracy of subsequent dissipation rate calculations, a phase alignment mechanism is designed within the module.

[0085] Because the sampling delay of electrical signals differs from the acquisition delay of mechanical signals, the module uses timestamps to interpolate and resample the two sets of data to ensure that the input active power and output mechanical power are strictly aligned at the same physical moment.

[0086] After obtaining the synchronized input active power and output mechanical power, the module enters the core calculation stage of energy dissipation rate.

[0087] According to the law of conservation of energy, part of the electrical energy input to the system is converted into effective mechanical work at the output, while the other part is dissipated in the form of heat, sound, etc. This dissipation includes the inherent losses of the equipment under normal conditions (such as stator copper loss, iron loss, and conventional frictional loss of bearings) as well as abnormal losses caused by deterioration of operating conditions or sudden changes in cycle time.

[0088] In order to accurately identify abnormal losses that reflect the physical limits of the equipment, the module is first calibrated under no-load or standard rated cycle time.

[0089] During the calibration phase, the module records the inherent losses at different speeds and temperatures, and fits them to generate a static inherent loss benchmark model.

[0090] In actual flexible production processes, the module calculates the dynamic energy dissipation rate in real time, using the following formula: ; in, The dynamic energy dissipation rate, For input active power, To output mechanical power, Current speed and temperature The static inherent loss is as follows; The formula calculates It has clear physical guiding significance. When the production line operates smoothly according to demand instructions or accelerates within a reasonable range, The value should fluctuate slightly around zero.

[0091] When demand commands require a significant increase in speed, causing resonance in the equipment's mechanical structure, rupture of the lubricating oil film leading to increased dry friction, or nonlinear deformation resistance of materials at high speeds, the input electrical power will increase dramatically. However, this increased power is not converted into effective mechanical output but is largely dissipated internally. The value will increase significantly.

[0092] Due to the complexity of industrial site conditions, the calculated original dynamic energy dissipation rate signal inevitably contains random noise caused by grid voltage fluctuations and instantaneous load disturbances.

[0093] If the original signal is used directly for subsequent control feedback, it can easily lead to malfunctions and frequent oscillations in the control system.

[0094] Therefore, this module must perform in-depth filtering and feature extraction on the output dissipation rate signal before it can output the signal.

[0095] The module employs a statistical algorithm based on Kalman filtering: This filtering process does not rely on a simple moving average, but is divided into two logical steps: prediction and update. In the prediction step, the filter calculates the predicted value of the dissipation rate at the current moment based on the optimal estimate of the dissipation rate at the previous moment and the state transition law of the system, and simultaneously calculates the covariance of the prediction error. In the update step, the filter reads the actual calculated dissipation rate observation with noise at the current moment, and dynamically calculates a Kalman gain coefficient by combining the prediction error covariance and the measurement noise covariance.

[0096] This gain coefficient determines whether the system trusts the predictions or the observations more. The filter uses this gain coefficient to weight and fuse the predictions and observations, outputting a smooth and optimal estimate that closely approximates the true physical state.

[0097] In addition, the module will perform trend analysis on the smoothed dissipation rate and extract its mean, variance and peak characteristics within a specific time window.

[0098] This energy dissipation rate data, which has undergone in-depth processing and feature extraction, will be stored in real time in the module's high-speed dual-port RAM and transmitted to the next module with extremely low latency via the internal high-speed backplane bus, serving as the core basis for determining whether the underlying physical device has reached its elastic adjustment limit.

[0099] The energy dissipation rate derivative constraint evaluation module is connected to the energy dissipation rate real-time calculation module. It is used to receive the dynamic energy dissipation rate after filtering and calculate the first-order time change rate. It generates a dynamic constraint threshold by combining the actual mechanical angular velocity and real-time temperature. The first-order time change rate is compared with the dynamic constraint threshold to calculate the risk index. The dissipation rate derivative constraint evaluation module is a key decision-making hub connecting the underlying physical state perception with the upper-level adaptive adjustment of the beat.

[0100] In Module 2, the system has already obtained a filtered estimate of the dynamic energy dissipation rate. In complex industrial production environments, a single absolute value of the dissipation rate is often insufficient to accurately determine whether the equipment is truly facing its physical limits.

[0101] For example, some heavy equipment has a high energy dissipation rate during normal acceleration; while for some precision equipment, even if the absolute value of the dissipation rate is not large, if its upward trend is too fast, it indicates that mechanical jamming or lubrication failure is about to occur.

[0102] Therefore, the core function of this module is no longer limited to static monitoring of the absolute value of the dissipation rate, but introduces dynamic evaluation of the trend of dissipation rate change (i.e., derivative). By calculating the first-order time rate of change of the energy dissipation rate at high frequency, the module can extremely sensitively capture abrupt changes in the nonlinear resistance inside the equipment.

[0103] Based on the current operating conditions of the equipment (such as speed and temperature), the module adaptively generates dynamic constraint thresholds and compares the real-time derivatives with these thresholds, ultimately outputting a quantified "physical limit approach risk index".

[0104] This index provides a forward-looking early warning signal for subsequent decentralized collaborative control, effectively preventing equipment from being damaged due to overload during the elastic speed-up process.

[0105] Calculating the derivative of a discrete signal presents a significant challenge in engineering practice. If the traditional two-point difference method is used (i.e., directly subtracting the value from the previous time step from the current value and then dividing by the sampling period), it will greatly amplify the high-frequency noise remaining in the signal, causing the calculated derivative signal to oscillate violently and completely lose its control reference value.

[0106] To address this issue, this module internally constructs a robust differentiator based on sliding window polynomial fitting. Instead of relying on simple two-point differences, this differentiator utilizes multiple historical data points over a period of time to reconstruct the smooth trajectory of the signal within a local window and calculates the analytical rate of change of that trajectory.

[0107] The specific calculation logic is as follows: The module maintains a fixed-length First-In-First-Out (FIFO) data queue in memory as a sliding time window. Whenever a new energy dissipation rate estimate is received, it is pushed onto the head of the queue, and the oldest data at the tail of the queue is pushed out.

[0108] The microprocessor inside the module extracts all data points within the window and uses time as the horizontal axis and dissipation rate as the vertical axis to find a quadratic parabolic trajectory that can most smoothly pass through these data points using the least squares method.

[0109] The core of the least squares method lies in continuously adjusting the coefficients of the constant, linear, and quadratic terms of the parabola so that the sum of the squares of the vertical distances from all actual data points to the parabola is minimized.

[0110] Once the coefficients of the optimally fitted parabola are determined, the microprocessor directly extracts the slope of the tangent line to the parabola at the latest moment. This tangent slope, which is the sum of the coefficients of the first term and twice the coefficients of the second term multiplied by the relative time, is the first derivative of the dissipation rate after high smoothing.

[0111] By using this polynomial fitting and differentiation method, the module not only obtains a high-precision rate of change of dissipation rate, but also fundamentally suppresses the amplification effect of high-frequency noise, ensuring the smoothness and reliability of the derivative signal.

[0112] After obtaining the smoothed dissipation rate derivative, it needs to be compared with a safety threshold. Traditional control systems typically set a fixed, hard threshold, but this is not feasible in flexible production.

[0113] Because the ability of a device to withstand sudden changes in dissipation rate is completely different when the device is running at low speed and high speed. Meanwhile, changes in ambient temperature or the temperature of the equipment itself can significantly alter the viscosity of the lubricating oil or the yield strength of the material, thereby affecting the safety threshold.

[0114] To address this, this module designs an adaptive generation mechanism for dynamic constraint thresholds; The module internally stores a set of physical characteristic surface models of the equipment based on a large amount of experimental data.

[0115] The model uses the current actual mechanical angular velocity of the equipment and the real-time temperature of key parts as independent variables to dynamically calculate the maximum allowable dissipation rate derivative threshold under the current operating conditions. The formula is: ; in, For the current moment Dynamic constraint threshold; The baseline dissipation rate derivative tolerance limit of the device when it is stationary and at standard room temperature; This represents the current actual mechanical angular velocity. The maximum speed designed for the physical operation of the equipment; This is the current real-time temperature. For standard operating temperature, The highest permissible temperature limit; and These are the speed penalty coefficient and the temperature penalty coefficient, respectively, used to characterize the degree to which increased speed and temperature weaken the equipment's tolerance.

[0116] This formula shows that as the equipment's rotational speed or temperature increases, its corresponding dynamic constraint threshold adaptively decreases. This mechanism ensures better protection for the equipment under harsh operating conditions, demonstrating a high degree of adaptability.

[0117] After calculating the dissipation rate derivative and dynamic constraint threshold in real time, the module enters the comprehensive evaluation stage; To avoid frequent changes in control commands caused by simple "black and white" logic judgments, the module introduces the idea of ​​fuzzy evaluation to calculate a continuously changing "physical limit approach risk index".

[0118] The risk index is a dimensionless floating-point number between 0 and 1.

[0119] 0 indicates that the equipment is operating extremely smoothly and is far from its physical limits; 1 indicates that the equipment has reached or exceeded its physical limits and must be stopped immediately or even slowed down urgently.

[0120] To achieve a smooth transition, the module sets a lower warning threshold, which is usually 60% of the dynamic constraint threshold.

[0121] The specific calculation logic of the risk index is as follows: The microprocessor first compares the derivative of the energy dissipation rate calculated in real time with the lower warning threshold. If the derivative is less than or equal to the lower threshold, it means that the energy dissipation rate of the current device is changing very slowly and is in an absolutely safe operating range. At this time, the microprocessor directly assigns a risk index value of 0.

[0122] If the derivative is greater than the lower threshold but less than the upper threshold of the dynamic constraint, it indicates that the dissipation rate is rising rapidly and the device is approaching its physical limit.

[0123] At this point, the microprocessor calculates the portion of the current derivative that exceeds the lower threshold and divides it by the difference between the upper and lower thresholds to obtain a basic linear approximation ratio.

[0124] To enhance the system's alertness when approaching its limits, the microprocessor amplifies this linear ratio nonlinearly (e.g., by raising it to the power of 1.5), causing the risk index to accelerate as it approaches its upper limit. If the derivative is greater than or equal to the upper limit threshold of the dynamic constraint, it indicates an extremely drastic change in the dissipation rate, reaching a dangerous state. In this case, the microprocessor will forcibly clamp the risk index to its maximum value of 1.

[0125] Through this segmented and nonlinear evaluation method, the module accurately quantifies the complex changes in the underlying physical state into a standardized risk index, providing a clear and continuous decision-making basis for subsequent control logic.

[0126] If the risk index is directly output after it is calculated, it may frequently switch between 0 and non-zero due to slight fluctuations in the signal under critical conditions, which may cause the entire production line control system to oscillate.

[0127] To eliminate this critical oscillation, this module incorporates a time-window-based hysteresis latch mechanism at the output.

[0128] Specifically, the module is equipped with a status monitoring timer. When the real-time calculated risk index changes from 0 to a non-zero value (i.e., an increase in risk is detected), the module will respond immediately and output the non-zero risk index to ensure the timeliness of the protection action.

[0129] However, when the risk index shows a downward trend or even drops to 0, the module does not immediately clear the value of the output port. Instead, it starts a decay timer. Within the time window set by the timer, the risk index of the output port will slowly decrease at a preset fixed slope until it decreases to 0. If the risk index rises again during the decrease, the timer is reset, and the output value follows the actual increase in the calculated value.

[0130] This hysteresis characteristic ensures that the system can remain in a "cautious" state for a period of time after encountering a sudden change in physical resistance, preventing premature acceleration from causing a secondary impact.

[0131] The risk index data packet, after being processed by hysteresis latch, will be tagged with a precise timestamp and the node ID of this process, and sent to the next module (reverse equivalent pressure transmission module) via the industrial Ethernet bus.

[0132] At the same time, the data packet will also be written into the module's internal non-volatile memory as a historical operation log for subsequent equipment health analysis and process optimization.

[0133] The reverse equivalent pressure transmission module, connected to the dissipation rate derivative constraint evaluation module, is used to receive the risk index and convert it into an initial reverse equivalent pressure through nonlinear mapping, generate the actual reverse equivalent pressure using the virtual impedance dynamics model, transmit the actual reverse equivalent pressure to the upstream process, and receive multi-source pressure signals transmitted from the downstream process and fuse them to generate a comprehensive reverse equivalent pressure. The reverse equivalent pressure transmission module is the core communication and dynamic conversion hub for achieving decentralized collaborative control and flexible adaptation. In traditional centralized control systems, if a bottleneck occurs in a certain process, the central controller usually issues a unified command to slow down the entire line. This approach not only has large communication delays but also easily leads to a "one-size-fits-all" approach that wastes production capacity.

[0134] This system adopts a decentralized architecture. When Module 3 assesses that the current process has a high risk of approaching the physical limit, the current process cannot simply rely on its own speed reduction to solve the problem, because this will cause the material transported from the upstream process to accumulate in front of the current process, eventually leading to material blockage or physical damage.

[0135] This module transforms the "risk index" of the current process into a virtual "reverse equivalent pressure" and simulates the physical characteristics of fluid dynamics or mechanical springs to transmit this pressure in reverse to the adjacent upstream process.

[0136] Upon receiving this pressure, the upstream process will spontaneously suppress its own production rhythm, thereby reducing the inflow of materials at the source. Through this pressure transmission mechanism based on a virtual physical field, the system achieves a flexible and adaptive balance of the entire production line's capacity through local interactions between adjacent nodes, without a central scheduling mechanism.

[0137] This module first receives the risk index from Module 3. In order to convert this dimensionless index into a control signal with a clear physical meaning, a nonlinear mapping function is constructed inside the module.

[0138] In the real physical world, when a pipe becomes blocked, the accumulation of reverse pressure often exhibits an exponential growth characteristic.

[0139] To realistically simulate this physical phenomenon, this module uses an exponential mapping formula to convert the risk index into an initial reverse equivalent pressure. The formula is: ; in, For initial reverse equivalent pressure (virtual unit: Pascal or equivalent force); The risk index received; The basic pressure gain coefficient determines the basic amplitude range of the pressure signal; The pressure expansion coefficient controls the steepness of the pressure increase with the risk index. Since the risk index ranges from [0,1], when the risk index is 0, the exponential term is 1, and subtracting 1 results in an initial reverse pressure of 0, indicating unimpeded downstream processes. As the risk index gradually increases, the initial reverse pressure spikes exponentially. This nonlinear mapping ensures that the system maintains efficient and flexible production under low risk conditions, while under high risk conditions, it can generate sufficiently strong "virtual resistance" to force upstream processes to slow down, thereby achieving reliable self-protection.

[0140] If the calculated initial reverse equivalent pressure is sent directly to the upstream process, the controller of the upstream process will receive a drastically changing step signal, which will also cause mechanical shock to the upstream equipment.

[0141] To ensure that the pressure transmission process conforms to real physical laws, this module constructs a "virtual impedance" dynamic model in the software logic.

[0142] The physical equivalent logic of this model is as follows: The module treats the virtual connection between two adjacent processes as a second-order mechanical vibration system containing a mass block, springs, and dampers. The calculated initial reverse equivalent pressure is considered as the external excitation force acting on this virtual mass block. The microprocessor inside the module uses a numerical integration algorithm to simulate the motion state of this virtual mass block under the action of the excitation force in real time.

[0143] Among them, the virtual mass coefficient determines the magnitude of the mass block's inertia. The greater the inertia, the slower the pressure signal changes, effectively filtering out high-frequency risk fluctuations. The virtual damping coefficient simulates the energy dissipation process and is used to absorb the oscillating energy during pressure transmission, preventing repeated speed fluctuations in upstream processes when receiving pressure. The virtual spring stiffness coefficient determines the magnitude of the actual transmitted pressure under steady-state conditions.

[0144] The microprocessor continuously calculates the displacement of the virtual mass block and multiplies this displacement by the virtual spring stiffness, ultimately obtaining a smooth, actual reverse equivalent pressure with physical inertia and damping characteristics. By introducing this virtual impedance model, the originally sharp, abrupt risk signal is transformed into a smooth pressure wave. This approach not only conforms to the dynamic constraints of the underlying equipment but also enhances the robustness of the decentralized control system.

[0145] After calculating the smoothed actual reverse equivalent pressure, this module needs to transmit it securely and on time to the corresponding module in the upstream process. In the decentralized architecture, communication between nodes relies on an industrial fieldbus.

[0146] Due to network bandwidth limitations and switch queuing mechanisms, random network delays are inevitable during data packet transmission.

[0147] If upstream processes directly use pressure signals with random delays for control, it will cause a shift in the control phase, leading to rhythm oscillations throughout the production line. To address this issue, this module appends a high-precision timestamp when sending pressure data packets.

[0148] When the upstream process's reverse equivalent pressure transmission module receives the data packet, it executes the following time alignment reconstruction logic: the upstream module extracts the sending timestamp from the data packet and compares it with its own local current time to calculate the actual transmission delay time experienced by the data packet in the network.

[0149] The upstream module extracts the first few pressure data points from its own historical cache and calculates the rate of change (i.e., slope) of the pressure signal at the time of transmission.

[0150] Then, the upstream module assumes that the pressure signal continues to change at that slope during this delay in network transmission.

[0151] Based on this assumption, the upstream module adds the product of the slope and the delay time to the received historical pressure value, thereby calculating (reconstructing) the actual pressure value at the current moment.

[0152] Through this timestamp-based phase compensation and reconstruction mechanism, upstream processes can "predict" the actual pressure state at the current moment, eliminating the adverse effects of network latency on decentralized collaborative control.

[0153] In complex production line topologies (such as those involving confluence, splitting, or parallel processes), an upstream process may simultaneously receive reverse equivalent pressure signals from multiple downstream processes.

[0154] For example, a main conveyor belt may simultaneously feed material to three different packaging machines. When all three packaging machines generate reverse pressure, the main conveyor belt's control module must rationally fuse these multi-source pressure signals. This module incorporates multi-source pressure fusion logic at the upstream process's receiving end. The specific fusion execution steps are as follows: The upstream module first performs the aforementioned time alignment reconstruction on all received downstream pressure signals to ensure that all signals are on the same time base.

[0155] The module assigns a preset weighting coefficient to each reconstructed pressure signal based on the technological importance and physical buffer capacity of each downstream process, and then performs a weighted multiplication operation. After obtaining a set of weighted pressure values, the module does not use a simple summation or averaging algorithm, but instead executes the maximum value envelope extraction logic.

[0156] The microprocessor compares these weighted stress values ​​one by one, selects the maximum value, and uses it as the final comprehensive reverse equivalent stress.

[0157] The use of maximum envelope logic is to ensure that upstream processes always respond to the most critical downstream bottlenecks, preventing serious local risks from being masked by averaging.

[0158] After calculating the comprehensive reverse equivalent pressure, the signal will be latched in the module's high-speed register and trigger a hardware interrupt signal to notify the next module (the decentralized beat adaptive module) to prepare to read the pressure value, thereby making substantial corrections to the initial target beat instruction.

[0159] The decentralized beat adaptive module is connected to the demand signal parsing and mapping module and the reverse equivalent pressure transmission module, respectively. It is used to receive the initial target beat command and the comprehensive reverse equivalent pressure, calculate the speed drop value using the virtual compliant impedance control algorithm, subtract the speed drop value from the initial target beat command to generate the dynamic synthetic beat command, and drive the physical actuator after performing boundary clamping on the dynamic synthetic beat command. The decentralized takt adaptive module is the final execution end and command hub of the entire flexible production control system.

[0160] In the preceding modules, the system has completed the mapping from macro-level demand to the initial target beat (Module 1), and calculated the energy dissipation rate (Module 2) and its derivative risk index (Module 3) through underlying physical state perception, thereby transforming the downstream physical limit risk into reverse equivalent pressure (Module 4).

[0161] The core task of this module is: The receiving module receives the "initial target beat command" and the "comprehensive reverse equivalent pressure" calculated by module four, and then constructs a virtual compliant dynamic model within the local controller. Using this model, the module can adaptively and dynamically attenuate and correct the initial target beat based on the magnitude of the pressure transmitted downstream, generating the final actual drive command sent to the physical actuator.

[0162] This module successfully found a dynamic balance between "meeting macro-level production capacity requirements" and "adhering to underlying physical limits," enabling each decentralized node to spontaneously achieve flexible coordination and safe self-adaptation of production rhythm without intervention from a central host computer.

[0163] When this module receives the combined reverse equivalent pressure from module four, it cannot simply use a linear proportional subtraction method to reduce the production cycle time, as this would cause abrupt changes in control commands and trigger oscillations in the mechanical system. To impart physical compliance characteristics to the underlying equipment, similar to springs and dampers, this module introduces a virtual compliance impedance control algorithm into its control logic.

[0164] This algorithm treats the cycle time adjustment process of the current operation as an equivalent of a moving mass block subjected to external resistance. The combined reverse equivalent pressure is considered as an external virtual resistance acting on this mass block.

[0165] The module calculates the required decay time (i.e., velocity drop) based on this resistance. To balance response to the absolute value of pressure and early suppression of pressure change trends, the velocity drop calculation employs a proportional-differential (PD) virtual admittance model, as shown in the following formula: ; in, The calculated speed drop value (i.e., the amount of beats that need to be actively reduced); The received comprehensive reverse equivalent pressure; The virtual compliance coefficient (admittance coefficient) characterizes the sensitivity of the current process to downstream pressure. The larger the value, the "softer" the current process is, and the greater the active speed reduction will be when subjected to the same reverse pressure. This is the virtual damping coefficient, used to respond to the rate of change of pressure; The physical meaning of this formula is as follows: when the downstream pressure rises sharply (i.e., the pressure derivative is positive and large), the differential damping term rapidly increases the velocity drop value, achieving a forward-looking and rapid deceleration to prevent further material accumulation; when the pressure tends to stabilize (the derivative approaches zero), the effect of the differential term weakens, and the system relies on the proportional compliance coefficient to maintain a steady-state deceleration amplitude. By solving this model, the module accurately transforms the abstract risk pressure signal into a specific physical quantity of cycle time decay.

[0166] After calculating the velocity drop value, the module needs to synthesize it with the initial target beat command. The specific synthesis logic is as follows: The microprocessor inside the module reads the smoothed initial target beat instruction issued by the module in each control cycle, and subtracts the speed drop value that was just calculated from it to obtain a preliminary corrected beat instruction.

[0167] This subtraction operation ensures that as long as there are risks and pressures downstream, the actual target cycle time of the current process will be lower than the cycle time of macro demand.

[0168] In actual production, the physical bottlenecks of downstream processes are often dynamic. When the energy dissipation rate of downstream processes returns to normal and the risk index drops to 0, the comprehensive reverse equivalent pressure will also dissipate, causing the speed drop value to rapidly approach 0.

[0169] At this point, if the initial correction cycle command is instantly restored to the initial target cycle, a huge positive speed jump will occur, causing a severe secondary impact on the mechanical transmission system of the current process.

[0170] To prevent this "stress release shock," this module employs an exponential smooth recovery mechanism. The execution logic of this smooth recovery mechanism is as follows: The microprocessor monitors the changing trend of the speed drop value in real time. When it detects that the speed drop value is in a decreasing phase (i.e., the downstream pressure is easing, allowing the current process to speed up), the microprocessor no longer directly outputs the initial correction cycle command, but instead starts a recovery program based on a first-order inertial filtering algorithm.

[0171] Within each discrete sampling period, the microprocessor first calculates the difference between the initial target beat at the current moment and the final output beat of the previous period. Then, the microprocessor extracts the recovery time constant preset in memory and, in conjunction with the current sampling period, calculates a smoothing coefficient between 0 and 1.

[0172] The microprocessor multiplies the calculated difference by the smoothing coefficient to obtain the beat recovery increment for the current cycle. The microprocessor then adds this recovery increment to the final output beat of the previous cycle, using it as the dynamic synthesized beat instruction for the current cycle.

[0173] In this process, the setting of the recovery time constant is crucial. The microprocessor dynamically calls this constant based on the physical moment of inertia and mechanical stiffness of the equipment in the current process.

[0174] For heavy equipment with extremely large rotational inertia, the microprocessor uses a larger time constant, which reduces the smoothing coefficient and the incremental recovery of each iteration, resulting in an extremely smooth cycle recovery process that presents a slowly rising exponential curve. For light, high-frequency equipment, a smaller time constant is used to ensure agile response.

[0175] This exponential smooth recovery mechanism ensures that after the equipment leaves the hazardous operating condition, it can safely and gently re-track the macroscopic demand rhythm with an acceleration consistent with its own dynamic characteristics.

[0176] Despite the complex compliant calculations and smoothing processes described above, the generated dynamic synthetic beat commands theoretically possess a high degree of adaptability. However, before they can be converted into pulse or analog signals for underlying drivers (such as frequency converters and servo amplifiers), they must undergo a final clamping protection at the absolute physical boundary.

[0177] The absolute physical limit parameters of the current process actuator are stored in the non-volatile memory (EEPROM) inside this module, including: the maximum permissible physical cycle time, the minimum permissible physical cycle time (to prevent the equipment from losing lubrication or resonating at extremely low speeds), and the maximum permissible physical acceleration.

[0178] The specific execution logic of boundary clamping is divided into two steps: amplitude clamping and rate of change clamping. Step 1 (Amplitude Clamping): The microprocessor reads the currently generated dynamic synthesized beat instruction and compares it with the fixed maximum allowable physical beat. If the instruction is greater than the maximum value, it is forcibly truncated and assigned the maximum value. The microprocessor then compares the truncated instruction with the minimum allowable physical beat. If the instruction is less than the minimum value, it is forcibly boosted and assigned the minimum value. This step ensures that the instruction is always within a safe absolute speed range. The second step (rate of change clamping): The microprocessor subtracts the actual output instruction of the previous control cycle from the current instruction after amplitude clamping to obtain the instruction change amount for a single cycle.

[0179] The microprocessor divides this change by the length of the control cycle to calculate the actual acceleration required by the current instruction.

[0180] The microprocessor compares the actual required acceleration with the fixed maximum permissible physical acceleration. If the positive value of the actual required acceleration exceeds the maximum permissible positive acceleration, the microprocessor discards the current instruction and instead uses the actual output instruction from the previous cycle, plus the product of the maximum permissible positive acceleration and the control cycle, as the new output instruction. Similarly, if the actual required negative acceleration (i.e., deceleration) exceeds the maximum permissible negative acceleration, the microprocessor will use the actual output instruction of the previous cycle as a basis, subtracting the product of the maximum permissible negative acceleration and the control cycle, as the new output instruction. If the actual required acceleration is within the safe range, the instruction after amplitude clamping will be output directly.

[0181] Through this logical boundary clamping, the system ensures that no matter how drastic the fluctuations in upper-level demand or how abnormal the downstream reverse pressure, the instructions issued to the physical actuators are always limited to the absolute physical envelope that the equipment can safely withstand, fundamentally preventing catastrophic equipment damage caused by control instructions exceeding limits. The clamped digital instructions are converted into 0-10V analog voltage signals by the module's internal digital-to-analog converter (DAC), or packaged into motion control messages via fieldbus protocols to directly drive the underlying motors.

[0182] This module's work does not end after sending the final instructions to the underlying driver. To maintain state transparency and consistency throughout the decentralized system, this module must broadcast the actual execution status of the current process to the entire network.

[0183] The communication subroutine within the module executes status packaging and broadcasting logic before the end of each control cycle. The microprocessor assembles the final tick instruction actually sent to the driver at the current moment, the energy dissipation rate risk index calculated for the current process, and the hardware flag indicating whether boundary clamp protection has been triggered into a standard status feedback data frame.

[0184] To ensure time consistency across the entire network, the microprocessor calls its internal hardware clock to add a high-precision timestamp to the data frame based on the IEEE 1588PTP protocol.

[0185] The data frame is sent to the network via multicast through the industrial Ethernet bus.

[0186] After receiving the status frame, the upstream process can use it to verify whether the reverse pressure it sent has been correctly responded to by the downstream. Once the global MES system or digital twin monitoring platform receives the status frames from all decentralized nodes, it can reconstruct the dynamic cycle distribution map and physical risk heat map of the entire production line in real time.

[0187] This underlying state broadcasting mechanism enables the decentralized control system to maintain a high degree of local autonomy while still providing a globally consistent data view to the upper layers, providing extremely valuable underlying physical operation data for subsequent process formulation optimization and preventive maintenance.

[0188] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0189] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely for distinguishing one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0190] It should be understood that in the various embodiments of this application, the sequence number of each process does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0191] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0192] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A demand-driven flexible production control system, characterized in that, include: The demand signal parsing and mapping module is used to receive macro demand signals from the upper-level information system, convert the macro demand signals into initial target beat instructions, and distribute the initial target beat instructions to each decentralized node after smoothing them. The real-time energy dissipation rate calculation module is used to synchronously collect the input active power and output mechanical power of the actuator, calculate the dynamic energy dissipation rate by combining it with the pre-calibrated static inherent loss, and perform filtering processing on the dynamic energy dissipation rate. The dissipation rate derivative constraint evaluation module is used to receive the dynamic energy dissipation rate after filtering and calculate the first-order time change rate. It combines the actual mechanical angular velocity and real-time temperature to generate a dynamic constraint threshold. The first-order time change rate is compared with the dynamic constraint threshold to calculate the risk index. The reverse equivalent pressure transmission module is used to receive the risk index and convert it into an initial reverse equivalent pressure through nonlinear mapping, and then transmit the actual reverse equivalent pressure to the upstream process. It also receives multi-source pressure signals transmitted from the downstream process and fuses them to generate a comprehensive reverse equivalent pressure. The decentralized beat adaptive module receives the initial target beat command and the comprehensive reverse equivalent pressure to calculate the speed drop value. It subtracts the speed drop value from the initial target beat command to generate a dynamic synthetic beat command. After boundary clamping the dynamic synthetic beat command, it drives the physical actuator.

2. The demand-driven flexible production control system according to claim 1, characterized in that, In the demand signal parsing and mapping module, the specific steps for converting macroscopic demand signals into initial target beat instructions are as follows: After receiving macro-level demand signals from the upper-level information system, the integrity of the macro-level demand signals is verified by using a cyclic redundancy check algorithm, and timestamps are extracted to remove out-of-order data packets. Extract the remaining target output and remaining delivery time window from the macro demand signal, divide the remaining target output by the remaining delivery time window and combine it with the dynamically updated comprehensive efficiency compensation coefficient to calculate the global theoretical target cycle time; The initial target physical cycle time is obtained by multiplying the global theoretical target cycle time by the mechanical transmission conversion coefficient of the corresponding process. The initial target physical beat is smoothed by using a command shaping filter based on a nonlinear S-shaped acceleration and deceleration curve. During the acceleration increase phase, the acceleration is increased with a constant jerk; during the uniform acceleration phase, the maximum acceleration is maintained; and during the acceleration decrease phase, the acceleration is decreased with a constant negative jerk, thereby generating continuous initial target beat commands.

3. The demand-driven flexible production control system according to claim 1, characterized in that, The real-time energy dissipation rate calculation module synchronously collects the input active power and output mechanical power of the actuator, and calculates the specific contents of the dynamic energy dissipation rate by combining it with the pre-calibrated static inherent losses. The instantaneous voltage and current of the three phases are obtained by voltage transformers and Hall current sensors. The instantaneous voltage and current of the three phases are multiplied and accumulated to obtain the instantaneous total power. The input active power is calculated by integrating and averaging the instantaneous total power within the sliding data window. The rotor position is obtained by using an optical encoder and the actual mechanical angular velocity is calculated differentially. Combined with the actual mechanical torque obtained by the torque sensor, the output mechanical power is calculated by multiplying the actual mechanical angular velocity and the actual mechanical torque. The formula for calculating the dynamic energy dissipation rate is as follows: ； in, The dynamic energy dissipation rate, For input active power, To output mechanical power, Current speed and temperature The static inherent loss.

4. The demand-driven flexible production control system according to claim 3, characterized in that, The process of filtering the dynamic energy dissipation rate is as follows: A statistical algorithm based on Kalman filtering is adopted. In the prediction step, the predicted value of the dissipation rate at the current time is calculated based on the final estimated value of the dissipation rate at the previous time step and the preset state transition law, and the prediction error covariance is calculated simultaneously. In the update step, the dynamic energy dissipation rate with noise calculated at the current moment is read as the observation value. The Kalman gain coefficient is dynamically calculated by combining the prediction error covariance and the measurement noise covariance. The Kalman gain coefficient is used to weight and fuse the predicted and observed values ​​of the dissipation rate to output a smooth final estimate. Trend analysis is performed on the final estimated value, and the mean, variance, and peak characteristics within the preset time window are extracted and stored in a high-speed dual-port random access memory.

5. The demand-driven flexible production control system according to claim 1, characterized in that, The dissipation rate derivative constraint evaluation module calculates the first-order time rate of change in the following way: A fixed-length first-in-first-out data queue is constructed in memory as a sliding time window. Whenever the filtered dynamic energy dissipation rate is received, it is pushed to the head of the queue and the oldest data at the tail of the queue is pushed out. Using time as the horizontal axis and dynamic energy dissipation rate as the vertical axis, the coefficients of the constant term, the first term, and the second term are continuously adjusted using the least squares method to find a smooth fitting parabola that passes through all data points within the sliding time window. The slope of the tangent line of the smooth fitting parabola at the latest moment is extracted as the first-order time rate of change to suppress the amplification effect of high-frequency noise.

6. The demand-driven flexible production control system according to claim 1, characterized in that, The process of generating a dynamic constraint threshold by combining the collected actual mechanical angular velocity and real-time temperature, and then calculating the risk index by comparing the first-order time rate of change with the dynamic constraint threshold is as follows: ； in, For the current moment Dynamic constraint threshold, The baseline dissipation rate derivative tolerance limit, This represents the current actual mechanical angular velocity. The highest speed designed for physical purposes. For real-time temperature, For standard operating temperature, The highest permissible temperature limit, This is the speed penalty coefficient. This is the temperature penalty coefficient; When calculating the risk index, the lower limit threshold for early warning is set as a fixed proportion of the dynamic constraint threshold. When the first-order time rate of change is less than or equal to the lower limit threshold for early warning, the risk index is assigned a value of zero. When the first-order time rate of change is greater than the lower warning threshold and less than the dynamic constraint threshold, the portion of the first-order time rate of change that exceeds the lower warning threshold is calculated and divided by the difference between the dynamic constraint threshold and the lower warning threshold to obtain the linear approximation ratio. The linear approximation ratio is then nonlinearly amplified to generate a risk index. When the first-order time rate of change is greater than or equal to the dynamic constraint threshold, the risk index is forcibly clamped to the maximum value of one.

7. The demand-driven flexible production control system according to claim 1, characterized in that, The reverse equivalent pressure transmission module converts the risk index into initial reverse equivalent pressure and generates actual reverse equivalent pressure, specifically including: The formula for calculating the initial reverse equivalent pressure is: ； in, For the initial reverse equivalent pressure, As a risk index, Based on the basic pressure gain coefficient, The coefficient of pressure expansion; The virtual connection between two adjacent processes is regarded as a second-order mechanical vibration system containing a mass block, a spring, and a damper. The initial reverse equivalent pressure is taken as the external excitation force acting on the virtual mass block. The virtual mass coefficient determines the inertia of the mass block to filter out high-frequency fluctuations, the virtual damping coefficient absorbs oscillation energy, and the virtual spring stiffness coefficient determines the magnitude of the steady-state transmitted pressure. The displacement of the virtual mass block is calculated by a numerical integration algorithm, and the displacement is multiplied by the virtual spring stiffness coefficient to generate a smooth actual reverse equivalent pressure.

8. The demand-driven flexible production control system according to claim 7, characterized in that, The specific steps for generating a comprehensive reverse equivalent pressure include: Extract the sending timestamp from the received multi-source pressure signal data packets transmitted by downstream processes, compare the sending timestamp with the local current time to calculate the actual transmission delay time, extract the pressure data points in the historical buffer to calculate the rate of change of the pressure signal at the time of transmission, add the received historical pressure value to the product of the rate of change and the actual transmission delay time, and deduce the real pressure state at the current time to complete the time alignment reconstruction. Based on the importance of downstream processes and physical buffer capacity, weight coefficients are assigned to the reconstructed multi-source pressure signals and weighted multiplication is performed. The maximum value envelope extraction logic is executed to compare and select the maximum value among the weighted pressure values ​​as the comprehensive reverse equivalent pressure.

9. The demand-driven flexible production control system according to claim 1, characterized in that, The decentralized beat-adaptive module calculates the speed drop value by including: The formula used is as follows: ； in, The speed drop value, To comprehensively address the reverse equivalent pressure, This is a virtual compliance coefficient. This is the virtual damping coefficient; Within each control cycle, the received initial target beat command is read, and the calculated speed drop value is subtracted from the initial target beat command to obtain the preliminary correction beat command, ensuring that the actual target beat of the current process is lower than the macro demand beat when there is risk pressure downstream.

10. The demand-driven flexible production control system according to claim 9, characterized in that, When generating dynamic synthetic beat commands, if the speed drop value is detected to be in a decreasing phase, an exponential smoothing recovery mechanism based on a first-order inertial filtering algorithm is activated. The difference between the initial target beat command at the current moment and the final output beat of the previous cycle is calculated. The smoothing coefficient is calculated by dynamically calling the recovery time constant based on the physical moment of inertia and mechanical stiffness. The difference is multiplied by the smoothing coefficient to obtain the beat recovery increment and accumulated to the final output beat of the previous cycle. Boundary clamping of dynamically synthesized beat instructions includes: In amplitude clamping, if the dynamic synthesized beat command is greater than the maximum allowed physical beat, it is forcibly truncated and assigned the maximum allowed physical beat; if it is less than the minimum allowed physical beat, it is forcibly boosted and assigned the minimum allowed physical beat. In rate-of-change clamping, the actual required acceleration is obtained by dividing the instruction change in a single cycle by the control cycle. If the actual required acceleration exceeds the maximum allowable physical acceleration range, the instruction that drives the physical actuator is generated based on the actual output instruction of the previous cycle and the maximum allowable physical acceleration.

Images