Automated coordinated regulation system for production flow in discrete manufacturing

Abstract

This invention relates to the field of discrete manufacturing process control technology, and discloses an automated collaborative adjustment system for discrete manufacturing processes, comprising: node control units distributed at process nodes and collaborative control calculation units; the collaborative control calculation unit converts the node running cycle into time axis distribution parameters in a time-series logic association model, calculates the target adjustment vector based on the parameters, and uses a feedforward shaping mechanism to superimpose compensation pulses on the output leading edge of the target adjustment vector, so that the adjustment signal breaks through the physical execution dead zone of the underlying drive execution unit. This invention smooths out the difference in dynamic impedance between the ideal control law and the mechanical actuator through signal shaping, realizes a hysteresis-free response to extremely small disturbances, blocks the nonlinear amplification of errors within a rigid time-series framework, and ensures stable convergence of the control loop.

Description

Technical Field

[0001] This invention relates to an automated collaborative adjustment system for discrete manufacturing production processes, belonging to the field of discrete manufacturing production process control technology. Background Technology

[0002] Current discrete manufacturing processes consist of multiple interconnected process nodes, each typically operating independently according to a preset cycle time. Materials and information flow between nodes in a fixed sequence, forming a production chain. This control method relies on each process node strictly adhering to a predefined process blueprint. In actual operation, the control logic of a single process node only focuses on local parameter feedback, matching its own physical parameters with static process thresholds, making it difficult to perceive the global state evolution of the production chain. When upstream nodes experience random disturbances, downstream nodes, constrained by isolated control domains, cannot detect upstream state shifts. This static and non-cooperative control logic causes minute time-series deviations to cascade and amplify between nodes, ultimately triggering production line shutdowns or material blockages. Conventional improvement approaches absorb disturbances by adding physical buffers or reducing the global baseline cycle time. However, these methods increase work-in-process inventory backlog and reduce overall output efficiency, and cannot eliminate the architectural mismatch caused by the diffusion of local static control domains and global dynamic disturbances.

[0003] Further analysis reveals that the core contradiction in existing technologies lies in the impedance mismatch between the ideal control signal at the mathematical level and the underlying mechanical actuator. When the control system outputs adjustment commands in response to minor disturbances, the drive device experiences execution lag due to static friction dead zones and electromagnetic inertial hysteresis, resulting in loss of the adjustment commands during physical conversion. Although the industry has attempted to improve information transmission speed by introducing high-speed communication networks, the industrial inertia of separating design and execution means that the control system still cannot directly converge physical disturbances within the parameter framework of the information processing space. This limitation in design philosophy makes it difficult to offset underlying physical oscillations in real time within the logical dimension, thus causing the system to exhibit instability when facing complex operating conditions. Chinese invention patent application CN117875664A discloses a multi-task sequential cooperative automatic... The existing control methods and devices, by constructing a quality characteristic parameter relationship model and using a target optimization model to adjust the time-dimensional sequence of manufacturing processes, are essentially macro-logical dimension scheduling optimization. The premise of this technology is that the underlying actuators can completely and losslessly reproduce the logical instructions. In the actual working conditions of discrete manufacturing, the underlying drive units, such as conveyor belt motors and assembly actuators, generally have static friction dead zones and electromagnetic induction lag. The aforementioned existing technologies only filter the sequence in the data space, ignoring the difference in dynamic impedance between the mathematical ideal control law and the coarse physical actuator. The instruction energy cannot penetrate the physical execution dead zone, resulting in the optimal solution having a response loop at the physical level. The control instructions and execution characteristics are essentially misaligned, making it difficult to offset the underlying physical oscillations in real time in the information processing space. The timing dead zone cannot be eliminated, and the feedback distortion is prone to inducing a systemic logical deadlock across the entire line.

[0004] Therefore, how to construct a cross-node dynamic collaborative adjustment mechanism to eliminate the timing logic dead zone between processes and suppress the cascading amplification of timing deviations under the constraints of the mechanical characteristics of physical actuators has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: An automated collaborative adjustment system for discrete manufacturing production processes, comprising:

[0006] Multiple node control units and collaborative control calculation units are distributed at process nodes; each node control unit is electrically connected to the collaborative control calculation unit.

[0007] The node control unit is used to collect physical status data of the corresponding process nodes;

[0008] The collaborative control solution unit is used to convert the operating cycle time of each process node into the time axis distribution parameters in the time-series logic association model, and execute the following steps: Step S101, calculate the synchronization deviation energy function of the time-series logic association model based on the time axis distribution parameters, and extract the time-series logic interference between process nodes; Step S102, retrieve the length of the logic queue to be processed corresponding to the process node, thereby setting the dynamic constraint boundary of the process cycle time, and calculate the optimal time axis distribution feature with the goal of minimizing the synchronization deviation energy function, so as to generate the target adjustment vector; Step S103, call the dead zone feedforward shaping submodule, obtain the static start-up energy threshold of the underlying drive execution unit, and perform waveform shaping on the target adjustment vector according to the static start-up energy threshold. By superimposing compensation pulses on the output leading edge of the target adjustment vector, the energy integral of the issued adjustment signal on the time axis meets the resistance overcoming requirement of the underlying drive execution unit, so that the transient power generated by the adjustment system breaks through the physical execution dead zone of the underlying drive execution unit; The collaborative control solution unit issues the shaped target adjustment vector to the corresponding node control unit.

[0009] Preferably, when the collaborative control solution unit executes step S101, it defines the logical coupling relationship of each process node as the node transfer weight in the temporal logic association model, and uses the Laplace operator to extract the structural feature value of the temporal logic association model in order to determine the system out-of-step risk degree that characterizes the synchronization deviation energy function.

[0010] Preferably, when the collaborative control solution unit executes step S102, it uses the length of the logic queue to be processed as a proportional gain factor of the dynamic constraint boundary; when the length of the logic queue to be processed exceeds the preset stacking threshold, it shrinks the adjustment tolerance window of the process cycle to suppress the cascading transmission of timing logic interference in the timing logic association model.

[0011] Preferably, when the collaborative control solution unit executes step S102, it uses the gradient descent algorithm to search for an adjustment path within the dynamic constraint boundary that makes the synchronization deviation energy function converge, and converts the time axis stretching of the time-series logic association model into the frequency control parameter in the target adjustment vector.

[0012] Preferably, the dead-zone feedforward shaping submodule calculates the amplitude of the shaped output signal. Follow these rules: ,in, The amplitude of the shaped output signal. Adjust the original magnitude of the target vector. The feedforward compensation amplitude is determined based on the static startup energy threshold.

[0013] Preferably, the dead-zone feedforward shaping submodule performs energy conservation constraints, and while increasing the feedforward compensation amplitude, it simultaneously reduces the duty cycle of the target adjustment vector in a single adjustment cycle to keep the total power consumption of a single adjustment action constant.

[0014] Preferably, when performing waveform shaping, the collaborative control calculation unit uses an exponential decay law to define the amplitude change characteristics of the compensation pulse over time, so that after the compensation pulse instantaneously breaks through the physical execution dead zone, it smoothly returns to the original level of the target adjustment vector.

[0015] Preferably, the collaborative control solution unit is used to monitor the rate of change of the timing logic interference quantity; when the rate of change exceeds the preset slope threshold within 10ms, the logic waiting delay of the downstream process node is adjusted in advance through dynamic constraint boundary.

[0016] Preferably, the underlying drive execution unit includes a conveyor belt drive motor and an assembly execution mechanism; the static start-up energy threshold is stored in the memory of the dead-zone feedforward shaping submodule and is corrected according to the deviation of the operating frequency of the underlying drive execution unit between 50Hz and 60Hz.

[0017] Preferably, the collaborative control solution unit further includes a system stability assessment module; the system stability assessment module is used to reconstruct the logical connection path of the timing logic association model when the synchronization deviation energy function exceeds a preset safety threshold, so as to isolate the process nodes that generate timing logic interference.

[0018] Compared with the prior art, the beneficial effects of the present invention are:

[0019] 1. In discrete manufacturing production processes, the node state perception unit synchronously collects the actual completion timestamp and instantaneous logic queue length to construct a dual information feedback mechanism that integrates historical execution performance and immediate capacity margin. The deviation extraction and transformation unit maps the temporal fluctuations between discrete processes into deviation vectors in the state space. Combined with the dynamic constraint boundary determined by the instantaneous logic queue length, this enables the control commands generated by the collaborative control solution unit to accurately perceive the instantaneous throughput capacity of downstream nodes. This cross-dimensional information coupling allows the system to avoid blindly pursuing the timing alignment of a single node when dealing with upstream disturbances. Instead, it actively calls upon the downstream logic cache redundancy to distribute the timing pressure, avoiding the secondary blocking risks caused by isolated adjustments in traditional control logic, and achieving a smooth transition of the entire production chain under dynamic disturbances.

[0020] 2. The dynamic weight evaluation submodule configured in the collaborative control solution unit realizes online monitoring of the performance degradation status of each physical node in the production line by calculating the real-time eigenvalue drift rate of the system state transition matrix. When a certain process node approaches the system stability boundary due to mechanical wear or changes in material properties, this submodule adjusts the distribution of control energy by reconstructing the state weight matrix online, adaptively shifting the adjustment center towards the vulnerable node. This control law reorganization mechanism based on the evolution of system dynamic characteristics enables the control system to logically evolve with the performance degradation of physical entities. Without changing the hardware physical configuration, the stable operation cycle of the entire line is extended through the dynamic focus of software logic, significantly enhancing the adaptability of the global control system to local physical degradation.

[0021] 3. The dead-zone feedforward shaping submodule embedded within the collaborative control solution unit deeply correlates the ideal control signal at the mathematical level with the physical start-up characteristics of the underlying drive device. For low-amplitude adjustment signals that cannot overcome mechanical friction or electromagnetic inertia due to minute timing deviations, this module, while maintaining the constraint of total adjustment energy conservation, superimposes compensation pulses with nonlinear leading-edge characteristics in the time domain, enabling the issued actual commands to instantaneously break through the physical execution dead zone. This software-level signal waveform shaping mechanism smooths out the dynamic impedance difference between the ideal control law and the coarse mechanical actuator, allowing the production line to achieve a hysteresis-free physical response to extremely minute disturbances at the 50ms level without using high-precision and expensive hardware. It also blocks the nonlinear amplification of small errors within a rigid timing framework, ensuring stable convergence of the control loop under complex operating conditions. Attached Figure Description

[0022] Figure 1 This is a flowchart of the discrete manufacturing process cycle time coordination adjustment and logic optimization of the present invention.

[0023] Figure 2 This is a schematic diagram of the distributed node collaborative control architecture and signal closed-loop interaction of the present invention.

[0024] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0025] The technical solutions of the embodiments of this application will be clearly described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of this application are within the scope of protection of this application.

[0026] An automated collaborative adjustment system for discrete manufacturing processes includes:

[0027] Multiple node control units and collaborative control calculation units are distributed at process nodes; each node control unit is electrically connected to the collaborative control calculation unit.

[0028] The node control unit is used to collect physical status data of the corresponding process nodes;

[0029] The collaborative control solution unit is used to convert the operating cycle time of each process node into the time axis distribution parameters in the time-series logic association model, and execute the following steps: Step S101, calculate the synchronization deviation energy function of the time-series logic association model based on the time axis distribution parameters, and extract the time-series logic interference between process nodes; Step S102, retrieve the length of the logic queue to be processed corresponding to the process node, thereby setting the dynamic constraint boundary of the process cycle time, and calculate the optimal time axis distribution feature with the goal of minimizing the synchronization deviation energy function, so as to generate the target adjustment vector; Step S103, call the dead zone feedforward shaping submodule, obtain the static start-up energy threshold of the underlying drive execution unit, and perform waveform shaping on the target adjustment vector according to the static start-up energy threshold. By superimposing compensation pulses on the output leading edge of the target adjustment vector, the energy integral of the issued adjustment signal on the time axis meets the resistance overcoming requirement of the underlying drive execution unit, so that the transient power generated by the adjustment system breaks through the physical execution dead zone of the underlying drive execution unit; The collaborative control solution unit issues the shaped target adjustment vector to the corresponding node control unit.

[0030] Preferably, when the collaborative control solution unit executes step S101, it defines the logical coupling relationship of each process node as the node transfer weight in the temporal logic association model, and uses the Laplace operator to extract the structural feature value of the temporal logic association model in order to determine the system out-of-step risk degree that characterizes the synchronization deviation energy function.

[0031] Preferably, when the collaborative control solution unit executes step S102, it uses the length of the logic queue to be processed as a proportional gain factor of the dynamic constraint boundary; when the length of the logic queue to be processed exceeds the preset stacking threshold, it shrinks the adjustment tolerance window of the process cycle to suppress the cascading transmission of timing logic interference in the timing logic association model.

[0032] Preferably, when the collaborative control solution unit executes step S102, it uses the gradient descent algorithm to search for an adjustment path within the dynamic constraint boundary that makes the synchronization deviation energy function converge, and converts the time axis stretching of the time-series logic association model into the frequency control parameter in the target adjustment vector.

[0033] Preferably, the dead-zone feedforward shaping submodule calculates the amplitude of the shaped output signal. Follow these rules: ,in, The amplitude of the shaped output signal. Adjust the original magnitude of the target vector. The feedforward compensation amplitude is determined based on the static startup energy threshold.

[0034] Preferably, the dead-zone feedforward shaping submodule performs energy conservation constraints, and while increasing the feedforward compensation amplitude, it simultaneously reduces the duty cycle of the target adjustment vector in a single adjustment cycle to keep the total power consumption of a single adjustment action constant.

[0035] Preferably, when performing waveform shaping, the collaborative control calculation unit uses an exponential decay law to define the amplitude change characteristics of the compensation pulse over time, so that after the compensation pulse instantaneously breaks through the physical execution dead zone, it smoothly returns to the original level of the target adjustment vector.

[0036] Preferably, the collaborative control solution unit is used to monitor the rate of change of the timing logic interference quantity; when the rate of change exceeds the preset slope threshold within 10ms, the logic waiting delay of the downstream process node is adjusted in advance through dynamic constraint boundary.

[0037] Preferably, the underlying drive execution unit includes a conveyor belt drive motor and an assembly execution mechanism; the static start-up energy threshold is stored in the memory of the dead-zone feedforward shaping submodule and is corrected according to the deviation of the operating frequency of the underlying drive execution unit between 50Hz and 60Hz.

[0038] Preferably, the collaborative control solution unit further includes a system stability assessment module; the system stability assessment module is used to reconstruct the logical connection path of the timing logic association model when the synchronization deviation energy function exceeds a preset safety threshold, so as to isolate the process nodes that generate timing logic interference.

[0039] Example 1: In a continuous material flow process consisting of multiple sets of conveyor belt drive motors and assembly execution mechanisms cascaded together, the production system experiences a 50ms-level high-frequency micro-time deviation caused by localized random mechanical friction at upstream process nodes. Conventional static cycle feedback adjustment architectures are limited by the static friction and electromagnetic inertial boundaries of the physical transmission components of the underlying assembly robot arm. When the original amplitude of the transient electrical signal of the control command sent to offset this micro-deviation is lower than the physical starting impedance of the underlying execution unit, the adjustment signal dissipates energy during the conversion phase. The upstream disturbance fails to receive a physical response at the target process node, causing nonlinear cascaded timing superposition errors to accumulate along the production link to subsequent nodes within the timing communication framework based on the inherent topology, triggering the timing logic interference of the entire line to exceed the limit and causing a shutdown interruption.

[0040] The node control unit collects the physical state data of the corresponding process node and extracts its actual completion timestamp. Simultaneously, it extracts the length of the pending logic queue for adjacent downstream process nodes. The collaborative control solution unit converts the operating cycle time of each process node into time axis distribution parameters in the time-series logic association model. Based on these time axis distribution parameters, it calculates the synchronization deviation energy function of the time-series logic association model, extracts the time-series logic interference between process nodes, and retrieves the length of the pending logic queue to set the dynamic constraint boundary of the process cycle time. The collaborative control solution unit calculates the optimal time axis distribution characteristics with the goal of minimizing the synchronization deviation energy function, generating the corresponding optimal control command vector U(k). This solution process follows the objective function... Where J is the synchronization deviation energy function, N is the prediction time-domain step number, i is the discrete prediction time step, and X(i) is the system state deviation vector mapped at the i-th step. Let X(i) be the transpose of X(i), Q be the state weight matrix in the temporal logic association model, and U(i) be the control command vector at step i. Let U(i) be the transpose of U(i), R be the control energy weight matrix, and k represent the current discrete time step. The collaborative control solution unit calls the dead-zone feedforward shaping submodule to obtain the static start-up energy threshold of the underlying drive execution unit. Based on this static start-up energy threshold, the waveform of the optimal control command vector U(k) is shaped, and the duty cycle of the adjustment signal within a single adjustment cycle is reduced synchronously according to the energy conservation constraint. The specific waveform shaping logic of the dead-zone feedforward shaping submodule is as follows: The original voltage amplitude of the optimal control command vector is compared with the static start-up energy threshold in real time. If the original voltage amplitude is less than 1.1 times the static start-up energy threshold, feedforward compensation is initiated. The calculation method for the feedforward compensation amplitude is: multiply the static start-up energy threshold by... A voltage redundancy coefficient of 1.25 is used as the peak voltage of the compensation pulse. The exponential decay rate parameter of the compensation pulse is calibrated through the following steps: Record the physical lag time between the actuator receiving the 1V step command and the encoder generating the first feedback pulse. If the lag time is 20ms, the compensation pulse is set to smoothly decay from the peak voltage to the original voltage level within 15ms. This ensures that the value of the pulse energy integral on the 10ms time axis is exactly equal to the energy value required to overcome the static friction of the mechanism, thereby instantaneously penetrating the physical execution dead zone of 1.2N·m. The compensation pulse, which exhibits exponential decay and has an amplitude exceeding the static start-up energy threshold, is superimposed on the output leading edge of the optimal control command vector U(k) to generate the target adjustment vector.

[0041] The collaborative control solution unit sends the shaped target adjustment vector to the corresponding node control unit. The energy integral of the adjustment signal on the time axis satisfies the resistance overcoming requirement of the underlying drive execution unit. The generated transient power breaks through the physical execution dead zone of the underlying drive execution unit, drives the assembly execution mechanism to respond and reduces the logical waiting delay of the downstream process node. The collaborative control solution unit monitors the rate of change of the timing logic interference quantity, and when the system stability assessment module determines that the synchronization deviation energy function exceeds the preset safety threshold, it reconstructs the logical connection path of the timing logic association model and isolates the process node that generates the timing logic interference quantity. Under the action of this composite control logic, the initial 50ms-level physical time difference deviation, under the condition of maintaining the original hardware topology arrangement, monotonically converges to the adjustment tolerance window of the process cycle within 3 to 5 control cycles. The physical execution dead zone variable of the underlying drive device is transformed into waveform shaping operation parameters inside the collaborative control solution unit.

[0042] Example 2: The test platform consists of a hardware-in-the-loop simulation system cascaded with a physical six-axis assembly robot arm, set to a continuous material flow condition. The platform is equipped with an incremental encoder with a resolution of 0.1ms to collect the actual completion timestamps of the physical joints. Simultaneously, the communication bus bandwidth of the servo driver is set to 2Mbps. To verify the signal processing status of the control architecture in an industrial electromagnetic environment, the signal generator injects Gaussian white noise with a signal-to-noise ratio of 20dB and power frequency electromagnetic harmonics at a frequency of 50Hz into the encoder feedback link. The discrete-time sampling period of the collaborative control solution unit is set to 1ms. This parameter setting is based on balancing the accuracy of transient physical friction capture with the bus communication load. Under the condition that the upper limit of the mechanical high-frequency jitter spectrum captured by the encoder approaches 500Hz, 1ms... The sampling period of s is based on the Nyquist sampling theorem to avoid signal aliasing in the discrete state space, thereby extracting micro-time deviations at the 50ms level. The experiment applies physical load mutations with equivalent timing delays of 20ms, 50ms and 80ms respectively to the target process node, and records and compares the response characteristics of different control architectures. The experiment is arranged into Comparative Sample Group 1, Comparative Sample Group 2 and the present invention sample group. Comparative Sample Group 1 adopts a proportional-integral-derivative closed-loop control law and has no dead-zone feedforward shaping mechanism. Comparative Sample Group 2 includes a dead-zone feedforward shaping submodule, and the prediction time-domain step N of its cooperative control solution unit is set to 2. The present invention sample group adopts a cooperative control solution unit, and its prediction time-domain step N is set to 10 according to the mapping relationship between the maximum buffer depth of the logic queue to be processed and the floating-point operation capability of the solver.

[0043] Under the condition of physical load mutation with an equivalent timing delay of 50ms, the actual completion timestamp fed back by the encoder exhibits a discrete random lag distribution. Compared with the sample group 1, the initial level amplitude of the output control command is lower than the equivalent voltage of the 1.2N·m physical static friction dead zone of the bottom assembly robot arm. The control energy dissipates in the early stage of electromagnetic conversion, and the assembly robot arm has no physical displacement output. The timing deviation accumulates to 145.5ms along the production link within 5 consecutive production cycles. Compared with the sample group 2, which is limited by the limited prediction time domain steps, its state deviation vector X(i) oscillates at high frequency. The output adjustment signal triggers the physical resonance of the robot arm. The node control unit of the present invention collects the actual completion timestamp containing Gaussian noise. Combined with the length of the unprocessed logic queue of the adjacent downstream related process node, the collaborative control solution unit calculates the synchronization deviation energy function. Solve for the optimal control command vector U(k), where J is the synchronization deviation energy function, N is the prediction time-domain step number, i is the discrete prediction time step, and X(i) is the system state deviation vector mapped at the i-th step. Let Q be the transpose of the corresponding system state deviation vector, Q be the state weight matrix in the time-series logic association model, and U(i) be the control command vector at step i. R is the transpose of the corresponding control command vector, R is the control energy weight matrix, and k represents the current discrete time step. The collaborative control solution unit identifies that the original amplitude of the optimal control command vector U(k) is lower than the static start-up energy threshold of the underlying drive execution unit. The dead-zone feedforward shaping submodule superimposes an exponentially decaying compensation pulse with an amplitude equivalent to 1.5 N·m on the leading edge of the waveform to generate the target adjustment vector. The monitoring equipment data shows that the transient power of the target adjustment vector breaks down the physical static friction impedance of 1.2 N·m. At the same time, the dynamic weight evaluation submodule in the collaborative control solution unit uses the Laplace operator to extract the real-time structural eigenvalue drift rate of the system state transition matrix and adaptively increases the corresponding node transfer weight in the state weight matrix Q. The timing deviation converges to 4.2 ms, and the dynamic adjustment of the state weight matrix Q suppresses the interference transmission of Gaussian white noise and power frequency harmonics.

[0044] In the verification of physical load mutations of different intensities, when the equivalent timing delay is 20ms, the timing deviation of the sample group of this invention converges to 1.8ms within 3 control cycles. When the equivalent timing delay is 80ms, the timing deviation converges to 12.5ms within 5 control cycles. When an over-limit disturbance with an equivalent timing delay of 120ms is injected into the test platform, the system response curve shows a nonlinear degradation inflection point. The transient voltage requirement of the compensation pulse approaches the hardware output saturation region of the servo amplifier. The target adjustment vector is affected by amplitude distortion, the convergence period of the timing deviation is extended to 25 control cycles, and the steady-state error increases to 38.4ms. The data gradient with physical saturation limit shows that the feedforward pulse energy adjustment mechanism based on quadratic objective function solution and waveform shaping has its adjustment capability boundary limited by the hardware output saturation of the underlying servo system. In the micro-time disturbance range of 20ms to 80ms, this mechanism can replace static friction hysteresis with waveform transient energy to offset the nonlinear timing superposition error between discrete production nodes without changing the underlying mechanical transmission topology.

[0045] Example 3: When the production system initially connects to a hardware topology consisting of multiple sets of conveyor belts and assembly actuators, there are dynamic modeling blind spots in the underlying physical transmission components caused by unknown mechanical friction. In order to obtain the static start-up energy threshold of the underlying drive actuator, the node control unit outputs a ramp step test voltage to the associated servo amplifier. This test voltage increases monotonically with a fixed time-domain step of 5mV. The node control unit simultaneously extracts the feedback position pulse of the assembly incremental encoder with a sampling period of 0.1ms. When the encoder first transmits a pulse level that represents the assembly robot arm generating an effective physical displacement of 0.05mm, the node control unit locks the transient drive voltage amplitude of the sampling period. The node control unit multiplies the transient drive voltage amplitude with the electromagnetic torque constant of the underlying drive actuator and sets the resulting product as the static start-up energy threshold of the current process node. The collaborative control calculation unit stores the static start-up energy threshold in the dead zone feedforward shaping submodule and transforms the static friction dead zone boundary of the underlying mechanical assembly components into the reference parameter for waveform shaping compensation calculation by the dead zone feedforward shaping submodule.

[0046] The collaborative control solution unit reads the physical material flow paths between each process node, maps the upstream and downstream connections to a directed graph topology in discrete space, and constructs an initial system state transition matrix A based on this directed graph topology. The collaborative control solution unit sets the state weight matrix Q in the temporal logic association model to a constant diagonal matrix with equally distributed diagonal elements. The collaborative control solution unit establishes a process node topology connection graph, defining adjacent process nodes with material flow relationships as directed graph edges. Weight values ​​are assigned to the elements of the adjacency matrix W according to the degree of material dependence between processes; a value of 1 is assigned if there is a direct material transport relationship between two processes, otherwise a value of 0 is assigned. System structural feature values ​​are extracted by calculating the Laplace matrix L=DW, where L is the Laplace matrix, D is the degree matrix, and W is the adjacency matrix. The full temporal logic interference vector is obtained by multiplying the Laplace matrix L with the local interference vector d. ,definition d is the temporal logic interference vector, and d is the local interference vector of a single node. The component values ​​represent the timing pressure exerted by specific process nodes on subsequent production links. In the cyclical calculation of continuous flow cycles, the dynamic weight evaluation submodule of the collaborative control solution unit obtains the system state transition matrix A updated in the current cycle and transforms it to generate the corresponding Laplace matrix. The dynamic weight evaluation submodule extracts the second smallest eigenvalue representing algebraic connectivity in the Laplace matrix as the structural eigenvalue and calculates the negative drift slope of the structural eigenvalue within adjacent time steps. When it is determined that the negative drift slope exceeds the system's set step loss risk tolerance threshold, the dynamic weight evaluation submodule analyzes the eigenvector sequence corresponding to the second smallest eigenvalue. The state weight evaluation submodule determines the abnormal process node that triggers the timing logic interference based on the index of the element with the largest absolute value in the feature vector sequence. The dynamic weight evaluation submodule extracts the diagonal element that maps to the abnormal process node in the state weight matrix Q and increases the value of the diagonal element according to the ratio of the negative drift slope. The timing logic association model reconstructs the solution trajectory of the synchronization deviation energy function J based on the adjusted state weight matrix Q, increases the weight of the error of the abnormal node, and drives the cooperative control solution unit to allocate the adjustment resources limited by the control energy weight matrix R to preferentially generate the control command vector for the abnormal process node.

[0047] Example 4: Before the control system is connected to the physical assembly line and put into trial operation, the node control unit executes the baseline calibration program. Based on the kinematic model of the six-axis assembly robot arm under no-load and set full-load conditions, a test instruction set is generated. The servo drive responds to the test instruction set in an environment without external material disturbance. The incremental encoder continuously collects the no-load and full-load displacement data of each joint. The deviation extraction and transformation unit calculates the time residual between the actual displacement data and the test instruction set. The statistical mean of the time residual is used as the systematic inherent bias of the reference process cycle. The collaborative control solution unit compensates the systematic inherent bias to the initial set value of the reference cycle and establishes sub-weight matrices for no-load and full-load conditions in the time logic association model. When the time logic association model identifies the change of the load type of the process node based on the production scheduling information, it synchronously calls the corresponding sub-weight matrix to form the solution basis of the synchronization deviation energy function J.

[0048] To determine the safety boundary for the risk of system out-of-synchronization, the collaborative control solution unit is configured with an adaptive calibration model. During the trial operation phase, it extracts the eigenvalue drift rate data that triggers production line shutdowns and calculates its probability density function. The collaborative control solution unit takes the upper limit of the drift rate corresponding to the 99.7% confidence interval of the probability density function as the initial value of the preset safety threshold. During continuous production cycles, the dynamic weight evaluation submodule collects the secondary eigenvalue drift rates that do not trigger shutdowns but cause secondary warnings in real time. The collaborative control solution unit uses a sliding time window method to periodically update the preset safety threshold. After a predetermined number of control cycles, the preset safety threshold is updated to the weighted average of the current initial value and the maximum value of the secondary eigenvalue drift rate within the sliding time window, so as to unify the mathematical solution with the dynamically changing physical production environment.

[0049] Example 5: When the system is deployed in an interactive operation between a conveyor belt with a material buffer station and an assembly actuator, the collaborative control calculation unit initiates a constraint calibration program. The node control unit extracts the depth dimension parameter of the physical buffer corresponding to the target process node, divides this depth dimension parameter by the physical geometric length of a single standard material unit, truncates it, and outputs the maximum permissible stacking base for that process node. The collaborative control solution unit obtains the depth parameters of the physical buffer zone of the target process node. Physical geometric length of standard material unit Calculate the maximum permitted stacking base ,definition For depth parameters, For physical geometric length, To determine the maximum permitted stacking base, according to the formula Obtain the calculated value and collect the length of the current node's pending logic queue. Substitute into the formula Calculate the dynamic scaling multiplier λ, define λ is the length of the logic queue to be processed, λ is the dynamic scaling multiplier, and the system preset baseline process tolerance time constant is extracted. According to the formula Calculate the dynamic constraint boundary of the current discrete time step process cycle. ,definition The baseline process tolerance time constant, As the dynamic constraint boundary of the process cycle, with As the gradient descent algorithm performs the optimal time axis distribution feature search for numerical constraint boundaries, it suppresses the cascading propagation of temporal logic interference within the temporal logic association model. During continuous flow cycles, the collaborative control solution unit collects the length of the unprocessed logic queue at the current node. According to the formula Calculate the dynamic scaling multiplier λ, where λ is the dynamic scaling multiplier. The length of the logic queue to be processed. To maximize the permitted stacking base, the collaborative control solution unit extracts the baseline process tolerance time constant from the system memory. According to the formula Calculate and output the dynamic constraint boundary of the process cycle corresponding to the current discrete time step. ,in This serves as a dynamic constraint boundary for the process cycle time. Using the baseline process tolerance time constant as a reference, the above algebraic operation process translates the physical material queuing state into continuous mathematical variables, which constitute the numerical constraint input for the time axis distribution feature optimization space.

[0050] Before the underlying drive execution unit is loaded and running, the collaborative control calculation unit triggers an offline waveform parameter test program. The node control unit applies a step change test voltage to the servo amplifier in a statically locked state, and simultaneously extracts the feedback reading of the incremental encoder at a sampling frequency of 0.1ms. It records the time span from the voltage change point to the encoder's first output pulse representing the mechanical shaft deflection. The dead-zone feedforward shaping submodule calibrates the static start-up energy threshold of the underlying drive execution unit. The node control unit outputs a monotonically increasing test voltage in 5mV steps to the servo driver, and simultaneously acquires the incremental encoder feedback pulse. It records the transient drive voltage amplitude when the encoder generates a displacement pulse and sets it as the static start-up energy threshold. ,definition The static start-up energy threshold is determined by the formula. > Determine the feedforward compensation amplitude ,definition Adjust the original magnitude of the target vector. To compensate for the feedforward amplitude, an exponentially decaying compensation pulse is superimposed on the leading edge of the optimal control command vector U(k). U(k) is defined as the optimal control command vector, and k is the discrete time step. The compensation pulse decay constant τ is adjusted so that the pulse energy integral meets the requirement of overcoming physical static friction. τ is defined as the decay constant. According to the principle of energy equivalence, the duty cycle of a single adjustment period is reduced to keep the total power consumption constant. The deviation extraction and transformation unit extracts this time span as the critical hysteresis constant of the corresponding physical transmission component. In the online timing waveform reconstruction operation, the dead zone feedforward shaping submodule calculates the reciprocal of the critical hysteresis constant and assigns its value as the exponential decay rate parameter of the compensation pulse. The transient power superimposed on the leading edge of the waveform decays synchronously to the basic amplitude level of the optimal control command vector according to the exponential decay rate after breaking through the physical execution dead zone.

[0051] Example 6: In an offline simulation environment, the system executes a weight sensitivity calibration program for the logical coupling relationship between different process nodes. By injecting a step timing deviation into a single node and observing the change amplitude of the full-chain synchronization deviation energy function J, the initial node transfer weight corresponding to that node is determined. The transfer weights of each node are used as diagonal elements to construct the initial state weight matrix Q in the timing logic association model. When the structural eigenvalue drift rate σ calculated by the dynamic weight evaluation submodule satisfies the judgment condition σ>δ, where δ is a preset safety threshold and σ is the structural eigenvalue drift rate, the system uses a mapping relationship... Calculate the weight increment ,in The weight increment is η, and the preset proportional adjustment coefficient is η. The element component of the j-th process node in the eigenvector corresponding to the eigenvalue of this structure is used to realize the corresponding element in the state weight matrix Q. The online reconstruction process transforms the abstract logical coupling strength into an algebraic evolution process with a deterministic computational path, eliminating the risk of empirical dependence in setting weight coefficients.

[0052] When determining the timing logic interference between process nodes, the deviation extraction and transformation unit extracts the phase distribution of each node's operating cycle time on the global time axis. The difference between the actual completion timestamps of any two adjacent process nodes is defined as the original timing interval. The absolute value of the difference between this original timing interval and the baseline process cycle time is calculated as the local interference of a single node. The collaborative control solution unit uses the Laplace operator to perform spatial correlation transformation on the total local interferometric quantities, and calculates the product of the adjacency matrix and the local interferometric vector to output the temporal logic interferometric vector characterizing the intensity of cascaded effects. ,in The vector represents the timing logic interference quantity. Each component of this vector corresponds to the timing constraint pressure exerted by a specific process node on the subsequent production chain. By mapping the physical deviation in the time dimension to the energy disturbance in the topological dimension, a quantified loss function input is provided for minimizing the synchronization deviation energy function J.

[0053] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0054] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An automated collaborative adjustment system for discrete manufacturing production processes, characterized in that, include: Multiple sets of node control units and collaborative control calculation units distributed at process nodes; Each node control unit is electrically connected to the collaborative control solution unit. The node control unit is used to collect physical status data of the corresponding process nodes; The collaborative control solution unit is used to convert the running cycle time of each process node into the time axis distribution parameters in the time-series logic association model, and executes the following steps: Step S101, calculate the synchronization deviation energy function of the time-series logic association model based on the time axis distribution parameters, and extract the time-series logic interference between process nodes; Step S102, retrieve the length of the logic queue to be processed corresponding to the process node, thereby setting the dynamic constraint boundary of the process cycle time, and calculate the optimal time axis distribution feature with the goal of minimizing the synchronization deviation energy function, so as to generate the target adjustment vector. In step S102, the collaborative control solution unit uses the length of the logic queue to be processed as the proportional gain factor of the dynamic constraint boundary; when the length of the logic queue to be processed exceeds the preset stacking threshold, the adjustment tolerance window of the process cycle time is shrunk to suppress the cascading transmission of time-series logic interference within the time-series logic association model; Step S103, call the dead-zone feedforward shaping submodule to obtain the bottom... The static startup energy threshold of the layer-driven execution unit is used to shape the waveform of the target adjustment vector. By superimposing a compensation pulse on the output leading edge of the target adjustment vector, the energy integral of the sent adjustment signal on the time axis meets the resistance overcoming requirement of the bottom-level drive execution unit, so that the transient power generated by the adjustment system can break through the physical execution dead zone of the bottom-level drive execution unit. The dead zone feedforward shaping submodule performs energy conservation constraints. When increasing the feedforward compensation amplitude, it simultaneously reduces the duty cycle of the target adjustment vector in a single adjustment cycle to keep the total power consumption of a single adjustment action constant. When the collaborative control solution unit performs waveform shaping, it uses an exponential decay law to define the amplitude change characteristics of the compensation pulse over time, so that after the compensation pulse breaks through the physical execution dead zone instantaneously, it smoothly returns to the original level of the target adjustment vector. The collaborative control solution unit sends the shaped target adjustment vector to the corresponding node control unit.

2. The automated collaborative adjustment system for discrete manufacturing production processes according to claim 1, characterized in that, When the collaborative control solution unit executes step S101, it defines the logical coupling relationship of each process node as the node transfer weight in the temporal logic association model, and uses the Laplace operator to extract the structural feature value of the temporal logic association model in order to determine the system out-of-step risk degree that characterizes the synchronization deviation energy function.

3. The automated collaborative adjustment system for discrete manufacturing production processes according to claim 1, characterized in that, When the collaborative control solution unit executes step S102, it uses the gradient descent algorithm to search for an adjustment path that makes the synchronization deviation energy function converge within the dynamic constraint boundary, and converts the time axis stretching of the time-series logic association model into the frequency control parameter in the target adjustment vector.

4. The automated collaborative adjustment system for discrete manufacturing production processes according to claim 1, characterized in that, The dead-zone feedforward shaping submodule calculates the amplitude of the shaped output signal. Follow these rules: ,in, The amplitude of the shaped output signal. Adjust the original magnitude of the target vector. The feedforward compensation amplitude is determined based on the static startup energy threshold.

5. The automated collaborative adjustment system for discrete manufacturing production processes according to claim 1, characterized in that, The collaborative control solution unit is used to monitor the rate of change of the timing logic interference quantity; when the rate of change exceeds the preset slope threshold within 10ms, it performs advance compensation adjustment on the logic waiting delay of the downstream process node through dynamic constraint boundary.

6. The automated collaborative adjustment system for discrete manufacturing production processes according to claim 1, characterized in that, The underlying drive execution unit includes a conveyor belt drive motor and an assembly execution mechanism; the static start-up energy threshold is stored in the memory of the dead-zone feedforward shaping submodule and is corrected according to the deviation of the underlying drive execution unit's operating frequency between 50Hz and 60Hz.

7. The automated collaborative adjustment system for discrete manufacturing production processes according to claim 1, characterized in that, The collaborative control solution unit also includes a system stability assessment module; the system stability assessment module is used to reconstruct the logical connection path of the timing logic association model when the synchronization deviation energy function exceeds the preset safety threshold, so as to isolate the process nodes that generate timing logic interference.

Images