Intelligent scheduling method for stereoscopic warehouse based on linkage digital twinning

Abstract

This invention relates to the field of digital twin warehouse control technology and discloses an intelligent scheduling method for a three-dimensional warehouse with linked digital twins. The method includes: acquiring the real-time coordinate sequence of physical work units and the virtual coordinates of the digital twin model, and calculating the virtual-physical synchronization deviation rate; determining that the work unit is in a response lag state when the deviation rate exceeds the lag threshold; and using the retrieval path reset mechanism of the three-dimensional topology database to intercept virtual node read requests for the work unit and implement node resampling. This method eliminates the risk of global scheduling clock suspension caused by the lag of a single entity by creating local virtual waiting at the logical level, ensuring the continuity of data flow of other work units in a concurrent environment, and improving the operational stability of the three-dimensional warehouse under high throughput conditions.

Description

Technical Field

[0001] This invention belongs to the field of digital twin warehouse control technology, and in particular relates to an intelligent scheduling method for a three-dimensional warehouse with linked digital twins. Background Technology

[0002] Currently, using digital twin models to construct the motion trajectory of execution units and deduce scheduling instructions based on this is the mainstream way to improve warehouse circulation efficiency. However, due to the interference of random factors such as mechanical wear, load fluctuations, and differences in driver response characteristics, the physical entities in the automated warehouse experience phase drift relative to the ideal timing of the virtual model. If the scheduling center issues control instructions based on the virtual state that has generated a leading deviation, it is easy to cause physical spatial interference in the multi-machine convergence area, leading to the system falling into an emergency stop deadlock state.

[0003] Traditional synchronization schemes mainly rely on adding high-precision sensors for position verification or truncating the global scheduling clock. However, the expansion of hardware scale significantly increases the system maintenance cost, and suspending the global clock forces adjacent concurrent units that have not experienced latency to stop running, resulting in severe throughput loss. The root cause lies in the fact that in existing twin systems built based on computer-aided design models, the retrieval logic of the underlying spatial topology database is rigidly bound to the global physical time axis, lacking the ability to decouple features for querying data streams of individual execution units. For example, Chinese invention patent CN114781841B discloses a digital twin generation... The production scheduling optimization method, device, equipment, and storage medium improves turnover efficiency by using a biomimetic swarm intelligence optimization algorithm to globally optimize the allocation of cargo locations and scheduling paths. This technology relies on the premise of highly linear synchronization between physical actions and virtual logic, but lacks dynamic fault tolerance for transient delays in individual work units. In high-throughput scenarios, there are differences in the material weight carried by work units and the dynamic response of motors. Individual lag is an objective law. Due to the lack of feature decoupling capability for individual data streams, once a node experiences a physical displacement deviation, the global optimization logic cannot absorb nonlinear disturbances in real time, resulting in misalignment between the scheduling instruction set and the physical action boundary, and inducing uncontrolled suspension of the global scheduling network.

[0004] Therefore, the technical problem to be solved by this invention is how to achieve local virtual-real phase alignment by optimizing the query logic of three-dimensional topology data while ensuring the continuity of global scheduling. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: A method for intelligent scheduling of a three-dimensional warehouse based on linked digital twins, comprising: Step S101: Obtain the real-time spatial coordinate sequence of the physical operation unit in the automated warehouse and the corresponding virtual node coordinates in the digital twin model, and obtain the motion inertia parameters of the physical operation unit and the preset sampling period of the digital twin model. Step S102: Based on the motion inertia parameters, calculate the virtual displacement increment of the virtual node coordinates within the current sampling period; based on the real-time spatial coordinate sequence, calculate the physical displacement increment of the physical operation unit within the current sampling period; calculate the absolute value of the difference between the virtual displacement increment and the physical displacement increment, and calculate the ratio of the absolute value of the difference to the sampling period to obtain the virtual-real synchronization deviation rate. Step S103: Compare the virtual-real synchronization deviation rate with a preset lag threshold, and determine that the physical operation unit is in a response lag state when the virtual-real synchronization deviation rate exceeds the lag threshold. In step S104, in response to the response lag state, an independent retrieval spin window is opened in the three-dimensional topology database corresponding to the digital twin model. By resetting the retrieval path to the three-dimensional topology database, the virtual node read request for the physical work unit is intercepted, and the virtual state of the physical work unit is resampled in the current topology node so that the data flow of the other concurrent work units in the three-dimensional topology database remains continuous.

[0006] Preferably, the operation of calculating the virtual-real synchronization deviation rate in step S102 includes: step S201, obtaining the real-time acceleration and velocity vector of the physical operation unit; step S202, combining the global clock stepping frequency of the digital twin model, predicting the ideal virtual displacement of the physical operation unit in the next sampling period by performing time-domain integration on the real-time acceleration and velocity vector; step S203, calculating the Euclidean distance change rate between the ideal virtual displacement and the actual displacement in the real-time spatial coordinate sequence, and defining the Euclidean distance change rate as the virtual-real synchronization deviation rate.

[0007] Preferably, the operation of opening an independent retrieval spin window in step S104 includes: step S301, allocating an independent cache area in the memory image of the 3D topology database as a local state mirror of the physical job unit; step S302, modifying the index logic of the database engine to decouple the topology node query request for the physical job unit from the global time axis and map it to the resampled sequence in the cache area.

[0008] Preferably, it further includes: step S501, real-time monitoring of the cumulative deviation value of the physical operation unit in the response lag state; step S502, when the cumulative deviation value falls back to the preset safe step range, canceling the retrieval path reset direction of the three-dimensional topology database, so that the query process for the physical operation unit is reconnected to the global physical clock of the digital twin model.

[0009] Preferably, the method further includes: step S601, calculating the physical spacing matrix between multiple concurrent operation units in the automated warehouse based on the real-time spatial coordinate sequence; step S602, identifying the interfering unit group in the confluence region according to the physical spacing matrix, and increasing the sampling priority of the interfering unit group in the three-dimensional topology database.

[0010] Preferably, the resampling operation in step S104 includes: step S701, maintaining the current coordinate node of the physical work unit in the virtual space; step S702, performing zero-order hold processing on the virtual state of the physical work unit when the global clock of the digital twin model steps, until the displacement synchronization signal fed back by the physical work unit is received.

[0011] Preferably, the method further includes: step S801, constructing a virtual coordinate prediction model based on kinematic extrapolation; step S802, when a transient interruption occurs in the real-time spatial coordinate sequence, generating a local predicted coordinate set through the virtual coordinate prediction model to maintain the calculation of the virtual-real synchronization deviation rate.

[0012] Preferably, the retrieval path reversal operation in the three-dimensional topology database specifically includes: step S901, locking the retrieval address of the physical operation unit in the current coordinate database; step S902, when the global scheduling clock advances to the next cycle, keeping the retrieval address pointing to the current topology node, and realizing local virtual waiting in the three-dimensional topology database.

[0013] Preferably, the method further includes: step S1001, calibrating the sensitivity of the hysteresis threshold according to the time-varying characteristics of the load of the physical work unit; step S1002, lowering the hysteresis threshold when the load increases to trigger the retrieval spin window, so that the virtual-real synchronization deviation rate is stabilized within the range of 0 to 0.15.

[0014] Compared with existing technologies, the intelligent scheduling method for automated warehouses linked to digital twins in this invention has the following advantages: 1. In the intelligent scheduling of automated warehouses, by constructing an independent data query spin window in the three-dimensional spatial topology database, the system intercepts virtual node read requests for specific execution units, causing the virtual state of the disturbed unit to be cyclically resampled in the current topology node, maintaining the data flow continuity of other concurrent execution units, and eliminating the risk of global scheduling clock suspension caused by the lag of a single entity. This localized data processing method enables each execution unit in a multi-machine collaborative environment to have independent data retrieval, avoids the spread of physical transient delays to the global scheduling network, and ensures the continuity of system operation under high throughput conditions.

[0015] 2. By relying on the dynamic quantization of the virtual and real phase drift rate and the reverse injection of the time delay compensation step, the system transforms the mechanical operation deviation of the physical entity into a virtual yielding action at the logical level, ensuring that the scheduling instruction set is always anchored to the real physical execution boundary. Without changing the underlying drive electrical parameters of the execution unit, this mechanism adjusts the stepping beat of the virtual space topology to make the time axis of the digital twin model actively match the execution beat of the physical entity, eliminating spatial interference and emergency stop deadlock faults caused by state advance prediction, and improving the path planning accuracy of the automated warehouse scheduling system.

[0016] 3. By combining kinematic inertial parameters and extrapolation prediction models, the system autonomously generates a local predicted coordinate set when the physical space coordinate flow is transiently interrupted. This maintains the continuous calculation of the virtual and real phase drift rates and the verification logic of the spin window, ensuring the anti-disturbance operation of the control compensation loop in the network fluctuation environment. This risk hedging strategy for the missing underlying coordinate flow enhances the fault tolerance of the scheduling center to the sensor network state, reduces the probability of systemic collapse caused by data loss, and realizes the robust output of warehouse control logic in complex electromagnetic environments. Attached Figure Description

[0017] Figure 1 This is the main flowchart of the intelligent scheduling of a three-dimensional warehouse based on the virtual-real synchronization deviation of the present invention; Figure 2 This is a diagram of the threshold adjustment and coordinate prediction mechanism for dynamic load feedback in this invention. Detailed Implementation

[0018] 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.

[0019] A method for intelligent scheduling of automated warehouses based on linked digital twins includes: Step S101: Obtain the real-time spatial coordinate sequence of the physical operation unit in the automated warehouse and the corresponding virtual node coordinates in the digital twin model, and obtain the motion inertia parameters of the physical operation unit and the preset sampling period of the digital twin model. Step S102: Based on the motion inertia parameters, calculate the virtual displacement increment of the virtual node coordinates within the current sampling period; based on the real-time spatial coordinate sequence, calculate the physical displacement increment of the physical operation unit within the current sampling period; calculate the absolute value of the difference between the virtual displacement increment and the physical displacement increment, and calculate the ratio of the absolute value of the difference to the sampling period to obtain the virtual-real synchronization deviation rate. Step S103: Compare the virtual-real synchronization deviation rate with a preset lag threshold, and determine that the physical operation unit is in a response lag state when the virtual-real synchronization deviation rate exceeds the lag threshold. In step S104, in response to the response lag state, an independent retrieval spin window is opened in the three-dimensional topology database corresponding to the digital twin model. By resetting the retrieval path to the three-dimensional topology database, the virtual node read request for the physical work unit is intercepted, and the virtual state of the physical work unit is resampled in the current topology node so that the data flow of the other concurrent work units in the three-dimensional topology database remains continuous.

[0020] Preferably, the operation of calculating the virtual-real synchronization deviation rate in step S102 includes: step S201, obtaining the real-time acceleration and velocity vector of the physical operation unit; step S202, combining the global clock stepping frequency of the digital twin model, predicting the ideal virtual displacement of the physical operation unit in the next sampling period by performing time-domain integration on the real-time acceleration and velocity vector; step S203, calculating the Euclidean distance change rate between the ideal virtual displacement and the actual displacement in the real-time spatial coordinate sequence, and defining the Euclidean distance change rate as the virtual-real synchronization deviation rate.

[0021] Preferably, the operation of opening an independent retrieval spin window in step S104 includes: step S301, allocating an independent cache area in the memory image of the 3D topology database as a local state mirror of the physical job unit; step S302, modifying the index logic of the database engine to decouple the topology node query request for the physical job unit from the global time axis and map it to the resampled sequence in the cache area.

[0022] Preferably, in step S102, the virtual-real synchronization deviation rate is calculated using the following formula: Where δ is the virtual-real synchronization deviation rate; This represents the virtual displacement increment; ΔT represents the physical displacement increment; ΔT represents the sampling period interval.

[0023] Preferably, it further includes: step S501, real-time monitoring of the cumulative deviation value of the physical operation unit in the response lag state; step S502, when the cumulative deviation value falls back to the preset safe step range, canceling the retrieval path reset direction of the three-dimensional topology database, so that the query process for the physical operation unit is reconnected to the global physical clock of the digital twin model.

[0024] Preferably, the method further includes: step S601, calculating the physical spacing matrix between multiple concurrent operation units in the automated warehouse based on the real-time spatial coordinate sequence; step S602, identifying the interfering unit group in the confluence region according to the physical spacing matrix, and increasing the sampling priority of the interfering unit group in the three-dimensional topology database.

[0025] Preferably, the resampling operation in step S104 includes: step S701, maintaining the current coordinate node of the physical work unit in the virtual space; step S702, performing zero-order hold processing on the virtual state of the physical work unit when the global clock of the digital twin model steps, until the displacement synchronization signal fed back by the physical work unit is received.

[0026] Preferably, the method further includes: step S801, constructing a virtual coordinate prediction model based on kinematic extrapolation; step S802, when a transient interruption occurs in the real-time spatial coordinate sequence, generating a local predicted coordinate set through the virtual coordinate prediction model to maintain the calculation of the virtual-real synchronization deviation rate.

[0027] Preferably, the retrieval path reversal operation in the three-dimensional topology database specifically includes: step S901, locking the retrieval address of the physical operation unit in the current coordinate database; step S902, when the global scheduling clock advances to the next cycle, keeping the retrieval address pointing to the current topology node, and realizing local virtual waiting in the three-dimensional topology database.

[0028] Preferably, the method further includes: step S1001, calibrating the sensitivity of the hysteresis threshold according to the time-varying characteristics of the load of the physical work unit; step S1002, lowering the hysteresis threshold when the load increases to trigger the retrieval spin window, so that the virtual-real synchronization deviation rate is stabilized within the range of 0 to 0.15.

[0029] Example 1: In an automated warehouse application scenario undertaking high-throughput concurrent scheduling tasks, the automated warehouse contains multiple physical operation units with multiple node confluence areas in space. Each physical operation unit steps the global state according to the preset sampling period of the digital twin model. When nonlinear mechanical damping fluctuations under full-load conditions cause a specific physical operation unit to lag behind its nominal kinematic equation in physical response, the synchronous scheduling based on the global clock suspension causes other normally operating physical operation units to experience retrieval deadlock and state query interruption when accessing the three-dimensional topology database, reducing global throughput efficiency. The digital twin intelligent scheduling method used in this example transforms the mechanical-level latency compensation into concurrent retrieval conflict handling for the three-dimensional topology database. The scheduling center obtains the real-time spatial coordinate sequence of the physical operation unit and the corresponding virtual node coordinates in the digital twin model, synchronously obtains the motion inertia parameters of the physical operation unit and the preset sampling period, calculates the virtual displacement increment of the virtual node coordinates in the current sampling period based on the motion inertia parameters, calculates the physical displacement increment of the physical operation unit in the current sampling period based on the real-time spatial coordinate sequence, and calculates it using the formula. The virtual-real synchronization deviation rate is calculated, where δ represents the virtual-real synchronization deviation rate. Characterizing virtual displacement increment, ΔT represents the physical displacement increment, and ΔT represents the interval duration of the sampling period.

[0030] When a physical work unit is under extreme load and the virtual-real synchronization deviation rate exceeds the lag threshold dynamically adjusted based on the time-varying characteristics of the load, the scheduling center determines that the physical work unit is in a lag state. It allocates an independent buffer in the memory image of the 3D topology database and, by modifying the database engine's index logic, decouples topology node query requests for that physical work unit from the global time axis. This forcibly intercepts the next virtual node read request for the physical work unit, causing its virtual state to be continuously resampled for the current topology node within the buffer. This mechanism allows the lagging physical work unit to gain local virtual waiting margin at the logical level, while the data flow of other concurrent work units that are not lagging remains continuous. This resolves the technical contradiction between single entity lag compensation and the continuity of global concurrent scheduling. Based on the sensor network fluctuation phenomenon present in industrial sites, the scheduling center incorporates virtual coordinates based on kinematic extrapolation. The predictive model autonomously generates a local predicted coordinate set to replace the missing physical coordinate input when a transient interruption of 50ms occurs in the real-time spatial coordinate sequence, maintaining continuous calculation of the virtual-real synchronization deviation rate. Simultaneously, the scheduling center monitors the cumulative deviation value of physical operation units under lag conditions. When the cumulative deviation value falls back to the preset safe step range, the retrieval path reset direction of the 3D topology database is canceled, allowing the query process for the physical operation unit to be reconnected to the global physical clock of the digital twin model. This ensures that the digital twin warehouse control network maintains parallel operation of multi-body collaborative computation and continuity of scheduling command output when facing nonlinear disturbances of the underlying action units. For the spatial trajectory of multi-axis linkage within the 3D topology space, single-dimensional displacement comparison causes spatial mapping deviation. After obtaining the virtual displacement increment and physical displacement increment, the scheduling center extracts the difference between each axial component in the 3D coordinate system and applies the formula... The rate of change of Euclidean distance in three-dimensional space is calculated as the virtual-real synchronization deviation rate, where... These represent the three components of the virtual displacement increment in the three-dimensional coordinate system. The three components of the physical displacement increment in the three-dimensional coordinate system are represented respectively. ΔT is the sampling period interval. The output is an objective deviation value with consistent dimensions. The multi-axis linkage spatial position drift is integrated into a single judgment benchmark to maintain the rigor of spatial status monitoring and the accuracy of triggering basis.

[0031] Example 2: In a physical simulation environment of an automated warehouse network architecture carrying high-density throughput indicators, the performance of a concurrent retrieval conflict handling scheme for a three-dimensional topology database under complex interference conditions is evaluated. A physical measurement and control platform containing fifty independent physical operation units is constructed based on a discrete event simulation kernel. A laser ranging sensor array with a measurement accuracy of 0.1mm and a sampling frequency of 100Hz is integrated to obtain real-time spatial coordinate sequences. Gaussian white noise with a signal-to-noise ratio of 20dB and 50Hz power frequency interference electromagnetic harmonics are superimposed on the input link to reproduce the underlying sensor fluctuations of the industrial network segment in the field. The preset sampling period ΔT of the digital twin model is set to balance the real-time performance of coordinate data acquisition with the concurrent processing load of the central control node. When the extreme load change rate of the physical operation unit increases, the state anti-aliasing distortion caused by high-speed kinematics deduction is blocked, and the preset sampling period ΔT is limited to approach the lower limit of the system's hardware interruption. Based on this decision logic, 10ms is selected as a typical sampling interval parameter in this test to generate the data basis corresponding to the objective working conditions.

[0032] To address the nonlinear delay in the raw coordinate data caused by environmental electromagnetic harmonic interference and mechanical transient damping, the test architecture included a control group using a global clock suspension synchronization scheme and an experimental group using a retrieval path reset scheme. During a two-hour full-load concurrent operation cycle, the acquisition end captured a transient noise point with an amplitude of 45.5mm in the raw sensor output. The experimental group used a front-end Kalman filter unit to remove this high-frequency noise and then extracted the true physical displacement increment. The virtual-real synchronization deviation rate δ, calculated according to the formula, climbed to 1.25 mm / ms at a certain moment, exceeding the lag threshold of 0.80 mm / ms set by the system based on the current extreme load. After the control group detected this lag state, it triggered the global clock interrupt mechanism, causing the other forty-nine normally operating concurrent job units to experience retrieval stagnation when accessing the 3D topology database. The global throughput plummeted from the nominal 1200 pallets per hour to 415 pallets per hour. After the experimental group determined that a single physical job unit was in a response lag state, it allocated an independent cache in the memory image and modified the index logic of the database engine to map the node query requests for the specific lag unit to the cache. The data flow of the other forty-nine concurrent job units was not disturbed, and the global throughput measured by the experimental group remained at 1185 pallets per hour, indicating that the cache reset mechanism eliminated the interference effect of physical lag on global scheduling.

[0033] A gradient test system was established for the intervention duration parameter of the virtual coordinate prediction model under transient interruption conditions. Test samples were extracted from three dimensions: intervention duration at the lower limit of 20ms, the median of 50ms, and the upper limit of 80ms. The cumulative deviation and action continuity of the system were measured under different interruption durations. The measurement data showed that when the intervention duration was set to 20ms, the prediction coordinate set changed too frequently, causing local index oscillations, and the system's concurrent error rate reached 4.2%. When the intervention duration was set to 50ms, the virtual node coordinates output by the extrapolation model and the subsequent recovered physical coordinate sequence achieved a smooth transition, the cumulative deviation value stabilized within the safe step range, and the system's concurrent error rate decreased to 0.1%. When the intervention duration was extended to the upper limit of 80ms, ... The accumulation of nonlinear mechanical damping causes the kinematic extrapolation model to diverge, resulting in a steep performance inflection point. Test records indicate that the predicted coordinate set under this condition deviates from the physical limit trajectory. The accumulated deviation value exceeds the safe step range and triggers three spatial topological interference warnings. This set of gradient verification data delineates the physical safety boundary of the prediction intervention time and establishes 50ms as the optimal working window for balancing computational continuity and extrapolation fidelity under this condition. Under the multiple constraints of nonlinear mechanical hysteresis and sensor link noise, the retrieval path reorientation mechanism of the three-dimensional topological database forcibly restricts the virtual waiting range of a single failed entity to an independent buffer area, avoiding query interruptions induced by global concurrent clock suspension and maintaining the parallel operation of the digital twin warehouse control network.

[0034] Example 3: In the continuously concurrent flow of the automated warehouse's bottom-level monitoring and control network, the dynamic load borne by the physical operation unit fluctuates with the real-time loading and unloading of palletized materials. When the digital twin system determines that the virtual-real synchronization deviation rate exceeds the limit, using a fixed hysteresis threshold will cause mechanical vibration under light load conditions to trigger an unreal response hysteresis state, and physical hysteresis under extreme heavy load conditions is easily missed as normal stepping. The mismatch between the dynamic load and the static judgment benchmark reduces the triggering accuracy of the concurrent retrieval path reset mechanism of the three-dimensional topology database. Within the synchronization period of acquiring the real-time spatial coordinate sequence of the physical operation unit, the scheduling center collects the real-time load voltage signal through the pressure sensor array at the load-bearing node of the physical operation unit chassis. The scheduling center converts the real-time load voltage signal into physical load-bearing mass and, based on the preset basic nominal threshold,... The dynamic load ratio μ of the current sampling and the equipment stiffness compensation coefficient K determine the hysteresis threshold for dynamic updates. The specific calculation logic is as follows: using the formula The hysteresis threshold is calculated. ,in, Characterizing the hysteresis threshold, The nominal threshold of the system under no-load conditions is represented by μ, which represents the ratio of physical load mass to nominal no-load mass, and K represents the stiffness compensation coefficient determined by the physical properties of the equipment.

[0035] When a physical operation unit causes a transient interruption in the real-time spatial coordinate sequence, the scheduling center activates a virtual coordinate prediction model. The scheduling center retrieves the historical coordinate increment set from the three consecutive sampling periods before the interruption as the reference input, calculates the second-order difference value of the historical coordinate increment set to obtain the transient physical acceleration, and multiplies the transient physical acceleration by the current dynamic load ratio μ to output a correction compensation amount. Based on this correction compensation amount, it superimposes the basic uniform kinematic vector to generate a local predicted coordinate set to replace the missing coordinates. Based on the underlying pressure sensor data, the hysteresis threshold is calculated in real time and the synchronous control method of extrapolating the predicted coordinates is used to enable the digital twin scheduling architecture to adjust the judgment boundary and predicted trajectory of physical hysteresis under fluctuating load conditions. During the alternating flow of light and heavy loads, the physical operation unit matches the actual mechanical damping state of the retrieval path reset triggering time for the three-dimensional topology database, maintaining the accuracy of physical spatial interference judgment in the multi-node confluence area.

[0036] Example 4: In the pre-commissioning calibration scenario of a multi-node convergence automated warehouse digital twin system, the scheduling center drives the physical operation unit to operate under a continuous mass gradient from no load to nominal ultimate load according to the standard test trajectory. The scheduling center synchronously acquires the calibration load voltage signal and physical space coordinate sequence through the pressure sensor array and laser rangefinder array at the chassis load-bearing nodes. Combined with the digital twin model, it synchronously deduces the corresponding virtual node coordinates. The scheduling center extracts the calibration physical displacement increment and calibration virtual displacement increment within the same preset sampling period and calculates the steady-state space deviation. The scheduling center divides the steady-state space deviation by the dynamic load ratio μ of the corresponding load gradient to calculate the flexible deformation parameter of a specific mechanical structure. Based on this, it constructs... A hardware feature mapping matrix containing multiple sets of stiffness compensation coefficients K is constructed, and the physical acceleration response sequence under test conditions is input into the initial state array of the virtual coordinate prediction model as the physical boundary reference for kinematic extrapolation. The hardware feature mapping matrix is ​​stored in the non-volatile memory of the scheduling center using a 10-row, 2-column lookup table structure. The first column records the load gradient values ​​from 0 kg to 1000 kg in 100 kg increments, and the second column records the corresponding stiffness compensation coefficients. Each coefficient is obtained by applying a standard counterweight to the equipment and measuring the flexible deformation of the mechanical structure under a 2000 mm extension displacement. The value range is calibrated between 0.85 and 1.15, which is used to linearly proportionally correct the physical drift caused by dynamic loads.

[0037] After the measurement and control network switches to online concurrent scheduling mode, the scheduling center queries the hardware feature mapping matrix based on the dynamic load ratio μ fed back by the pressure sensor array in real time, extracts the stiffness compensation coefficient K corresponding to the current load state, and substitutes the stiffness compensation coefficient K into the hysteresis threshold. In the dynamic computing nodes, the virtual coordinate prediction model synchronously retrieves the physical boundary reference in the initial state array to verify the transient physical acceleration calculated in real time. The system relies on the inherent mechanical bias data of the physical platform to calibrate the trigger baseline and spatial extrapolation coordinate sequence of the three-dimensional topology database retrieval path reset mechanism.

[0038] Example 5: In the scenario of pre-parameter calibration for the physical flow of a multi-node converged automated warehouse digital twin system, the scheduling center controls the physical operation unit to continuously operate at the nominal maximum speed along a standard test track under no-load conditions. The laser ranging sensor array collects the physical displacement sequence, and the digital twin model synchronously outputs the virtual displacement sequence. The scheduling center calculates the time mean of the absolute difference between the physical displacement and the virtual displacement within a preset time window, and adds this time mean to the inherent sensing range to generate the basic nominal threshold of the system under no-load conditions. The scheduling center loads a standard weight array at the chassis load-bearing node of the physical operation unit to construct five equidistant mass gradients spanning from no load to full load. The steady-state space deviation is measured at each mass gradient. The scheduling center uses the least squares method to fit a linear mapping function between the steady-state space deviation and the corresponding dynamic load ratio μ. The inverse of the slope of the linear mapping function is extracted to establish the stiffness compensation coefficient K, thereby constructing the numerical architecture of the underlying judgment benchmark parameters.

[0039] When a transient interruption occurs in the real-time spatial coordinate sequence, triggering the system's extrapolation operation, the input port of the virtual coordinate prediction model, which has a built-in first-order autoregressive moving average architecture, receives the historical coordinate increment set within the three consecutive sampling periods prior to the interruption. The core logic component solves the second-order difference of the historical coordinate increment set to extract the transient physical acceleration. The transient physical acceleration is then introduced into a weighted average filter with a sliding window width of 10 sampling periods to filter out the high-frequency mechanical vibration envelope. The smoothed acceleration parameter output after filtering is multiplied by the current dynamic load ratio μ to generate a correction compensation amount. The output port outputs a local predicted coordinate set based on the vector sum and integral of this correction compensation amount and the basic uniform kinematic vector. By relying on the historical data difference and smoothing calculation path, the transient physical acceleration is constrained within the mechanically permissible envelope range, maintaining the physical spatial adaptability of the extrapolated coordinate sequence under environmental disturbances.

[0040] The embodiments of this application have been described above with reference to the accompanying drawings. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. This application is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of this application without departing from the spirit of this application and the scope of protection of this invention, and all of these forms are within the protection scope of this application.

Claims

1. A linkage digital twin stereoscopic warehouse intelligent scheduling method, characterized in that, include: Step S101: Obtain the real-time spatial coordinate sequence of the physical operation unit in the automated warehouse and the corresponding virtual node coordinates in the digital twin model, and obtain the motion inertia parameters of the physical operation unit and the preset sampling period of the digital twin model. Step S102: Based on the motion inertia parameters, calculate the virtual displacement increment of the virtual node coordinates within the current sampling period; Based on the real-time spatial coordinate sequence, calculate the physical displacement increment of the physical operation unit within the current sampling period; Calculate the absolute value of the difference between the virtual displacement increment and the physical displacement increment, and calculate the ratio of the absolute value of the difference to the sampling period to obtain the virtual-physical synchronization deviation rate; Step S103: Compare the virtual-real synchronization deviation rate with a preset lag threshold, and determine that the physical operation unit is in a response lag state when the virtual-real synchronization deviation rate exceeds the lag threshold. In step S104, in response to the response lag state, an independent retrieval spin window is opened in the three-dimensional topology database corresponding to the digital twin model. By resetting the retrieval path to the three-dimensional topology database, the virtual node read request for the physical work unit is intercepted, and the virtual state of the physical work unit is resampled in the current topology node so that the data flow of the other concurrent work units in the three-dimensional topology database remains continuous.

2. The linkage digital twin-based intelligent scheduling method for a stereoscopic warehouse according to claim 1, characterized in that, The operation of calculating the virtual-real synchronization deviation rate in step S102 includes: step S201, obtaining the real-time acceleration and velocity vector of the physical operation unit; step S202, combining the global clock step frequency of the digital twin model, predicting the ideal virtual displacement of the physical operation unit in the next sampling period by performing time-domain integration on the real-time acceleration and velocity vector; step S203, calculating the Euclidean distance change rate between the ideal virtual displacement and the actual displacement in the real-time spatial coordinate sequence, and defining the Euclidean distance change rate as the virtual-real synchronization deviation rate.

3. The linkage digital twin-based intelligent scheduling method for a stereoscopic warehouse according to claim 1, characterized in that, The operation of opening an independent retrieval spin window in step S104 includes: step S301, allocating an independent cache area in the memory image of the 3D topology database as a local state mirror of the physical job unit; step S302, modifying the index logic of the database engine to decouple the topology node query request for the physical job unit from the global time axis and map it to the resampled sequence in the cache area.

4. The linkage digital twin-based intelligent scheduling method for a stereoscopic warehouse according to claim 1, characterized in that, It also includes: step S501, real-time monitoring of the cumulative deviation value of the physical operation unit in the response lag state; step S502, when the cumulative deviation value falls back to the preset safe step range, canceling the retrieval path reset direction of the three-dimensional topology database, so that the query process for the physical operation unit is reconnected to the global physical clock of the digital twin model.

5. The intelligent scheduling method for a three-dimensional warehouse based on a linked digital twin according to claim 1, characterized in that, It also includes: step S601, calculating the physical spacing matrix between multiple concurrent operation units in the automated warehouse based on the real-time spatial coordinate sequence; step S602, identifying the interfering unit group in the confluence region according to the physical spacing matrix, and increasing the sampling priority of the interfering unit group in the three-dimensional topology database.

6. The intelligent scheduling method for a three-dimensional warehouse based on a linked digital twin according to claim 1, characterized in that, The resampling operation in step S104 includes: step S701, maintaining the current coordinate node of the physical work unit in the virtual space; step S702, when the global clock of the digital twin model steps, performing zero-order hold processing on the virtual state of the physical work unit until the displacement synchronization signal fed back by the physical work unit is received.

7. The intelligent scheduling method for a three-dimensional warehouse based on a linked digital twin according to claim 1, characterized in that, It also includes: step S801, constructing a virtual coordinate prediction model based on kinematic extrapolation; step S802, when a transient interruption occurs in the real-time spatial coordinate sequence, generating a local predicted coordinate set through the virtual coordinate prediction model to maintain the calculation of the virtual-real synchronization deviation rate.

8. The intelligent scheduling method for a three-dimensional warehouse based on a linked digital twin according to claim 1, characterized in that, The retrieval path reset operation in the 3D topology database specifically includes: step S901, locking the retrieval address of the physical operation unit in the current coordinate database; step S902, when the global scheduling clock advances to the next cycle, keeping the retrieval address pointing to the current topology node, and realizing local virtual waiting in the 3D topology database.

9. The intelligent scheduling method for a three-dimensional warehouse based on a linked digital twin according to claim 1, characterized in that, It also includes: step S1001, calibrating the sensitivity of the hysteresis threshold according to the time-varying characteristics of the load of the physical work unit; step S1002, lowering the hysteresis threshold when the load increases to trigger the retrieval spin window and stabilize the virtual-real synchronization deviation rate within the range of 0 to 0.15.

Images