A parcel sorting system and a control method thereof

Abstract

The present application belongs to the technical field of parcel sorting, and aims to provide a parcel sorting system and a control method thereof, and solve the poor adaptability of the existing cross-belt sorting machine to soft packages and irregular-shaped packages; the parcel sorting system comprises a parcel information acquisition unit and a multi-axis servo execution unit, the multi-axis servo execution unit comprises a plurality of servo motor drivers arranged in an array and a sorting execution mechanism connected with each driver in correspondence, further comprises a material flexibility grade quantization and division unit, a differential control instruction generation unit, a parcel ID binding and tracking unit and a microsecond-level control instruction scheduling hub.

Description

Technical Field

[0001] This invention belongs to the field of parcel sorting technology, and particularly relates to a parcel sorting system and its control method. Background Technology

[0002] With the rapid development of the express delivery industry, the total number and types of parcels are constantly increasing, and the proportion of irregularly shaped parcels and soft parcels is also increasing. Cross-belt sorting machines have become the core sorting equipment in express distribution centers due to their high sorting efficiency and wide applicability.

[0003] However, the core control parameters of existing cross-belt sorting equipment, such as acceleration / deceleration curves and unloading angles, are all preset fixed values. They do not take into account the differences in physical characteristics between soft packages and irregularly shaped packages. For soft packages, the packaging material has a low elastic modulus and large deformation, and the dynamic friction coefficient with the cross-belt surface fluctuates greatly. A fixed acceleration can easily cause the package to slide relative to the belt surface, resulting in a deviation between the unloading start position and the preset position, directly causing missorting. At the same time, soft packages have weak impact resistance, and the step impact of fixed acceleration / deceleration can easily cause damage to the package packaging and internal parts. Artificially reducing the acceleration will cause a significant drop in sorting efficiency. For irregularly shaped packages, their irregular shape, large center of gravity offset, and unstable contact area with the belt surface can easily cause packages to tip over, uneven force on the belt surface, and unloading trajectory deviation from the preset path under fixed control parameters. This can lead to missorting and missed sorting at best, and sorting line jamming and equipment shutdown at worst.

[0004] Therefore, developing a parcel sorting system and control method that can adapt to irregularly shaped flexible parcels and has microsecond-level precise scheduling capabilities has become an urgent technical problem to be solved in this field. Summary of the Invention

[0005] The purpose of this invention is to provide a parcel sorting system and its control method in order to solve the problem of poor adaptability of existing cross-belt sorting machines to soft and irregularly shaped parcels.

[0006] The technical solution includes a package information acquisition unit and a multi-axis servo execution unit. The multi-axis servo execution unit includes several servo motor drivers deployed in an array and sorting execution mechanisms connected to each driver. It also includes a material flexibility level quantification and classification unit, a differentiated control instruction generation unit, a package ID binding and tracking unit, and a microsecond-level control instruction scheduling center. The input end of the material flexibility level quantification unit is connected to the output end of the package information collection unit. It has a built-in standardized package material flexibility grading model to receive package feature parameters, complete the material flexibility level quantification of the package to be sorted, and output the corresponding package flexibility level label. The differentiated control command generation unit is communicatively connected to the hierarchical division unit. It has a built-in flexibility level-sorting control parameter mapping library and an adaptive S-shaped acceleration / deceleration algorithm model and unloading angle trajectory inversion model based on the flexibility level. It is used to match and generate corresponding package-specific acceleration curve segmentation control commands and unloading angle adjustment commands according to the received level labels. The algorithm model uses the package flexibility level as the core correction factor to achieve adaptive quantitative solution of acceleration, acceleration, acceleration segment duration and unloading angle. The package ID binding and tracking unit communicates bidirectionally with the acquisition unit and the scheduling center, respectively. It is used to assign a unique global ID to each package, complete the binding of package information, control instructions and global ID, track the package location in real time and update the mapping table of ID and corresponding execution mechanism and real-time location. The microsecond-level control command scheduling center is communicatively connected to the command generation unit, the tracking unit, and each servo driver. It adopts a hard real-time architecture with a command scheduling cycle of ≤100μs and a multi-axis synchronization error of ≤1μs. It is used to receive control commands and, based on the mapping table, precisely synchronize and schedule the commands to the corresponding servo drivers within the microsecond-level time window when the package enters the target workstation, thereby driving the actuator to complete the precise sorting of the package.

[0007] Preferably, the package information acquisition unit includes a synchronously triggered 3D laser contour scanning module, a hyperspectral material recognition module, a weighing module, and a high-speed vision module. The timestamp alignment error of the data acquired by each module is ≤100μs. It is used to acquire the elastic modulus, deformation coefficient, weight, dimensions, and irregular features of the package material. The hyperspectral material recognition module has a built-in material mechanical parameter inversion model, which can extract the elastic modulus of the package material based on hyperspectral reflectance data.

[0008] Preferably, the standardized flexibility grading model uses the material's elastic modulus and deformation coefficient as core grading indicators, and weight and irregularity as correction indicators. The package is divided into 5 flexibility levels through a grading decision tree. The higher the level number, the higher the package's flexibility and the weaker its impact resistance.

[0009] Preferably, the flexibility level-sorting control parameter mapping library matches corresponding control strategies based on the sorting failure modes of packages with different flexibility levels. High-flexibility packages are matched with parameters for low-impact S-shaped segmented acceleration and deceleration and reduced unloading angle, rigid packages are matched with parameters for high-response segmented acceleration and deceleration and increased unloading angle, and semi-rigid or semi-flexible packages are matched with balanced control parameters.

[0010] Preferably, the microsecond-level control instruction scheduling center adopts a hard real-time architecture of FPGA + EtherCAT industrial Ethernet bus, and realizes full system timing synchronization based on EtherCAT distributed clock, with an instruction end-to-end issuance delay of ≤50μs.

[0011] Preferably, the system further includes a sorting result feedback and closed-loop optimization unit. The unit includes a detection module deployed at the sorting grid, used to collect package sorting results, optimize corresponding control parameters based on missorted and damaged abnormal data, and optimize the grading model and parameter mapping library based on full sorting data.

[0012] A control method for a parcel sorting system includes the following steps: S1 collects full quantitative feature information of packages to be sorted, assigns a unique global ID to the package, and completes information binding; S2 inputs the feature information into the preset standardized flexibility grading model, completes the quantitative division of the flexibility level of the wrapping material, generates the corresponding level label and binds it to the global ID; S3 is based on a preset flexibility level-sorting control parameter mapping library and an adaptive algorithm model. It matches the level label to solve the segmented parameters of the package's specific acceleration curve and the optimal unloading angle, and generates corresponding control commands. S4 tracks the package location in real time and updates the mapping table between the global ID and the corresponding sorting execution mechanism and real-time location. Through the microsecond-level control command scheduling center, the control commands are precisely and synchronously scheduled to the corresponding servo motor driver within the microsecond-level time window when the package enters the target sorting station. The S5 servo motor driver receives instructions and drives the sorting execution mechanism to complete the segmented accelerated conveying, unloading angle adjustment and precise sorting of packages, while simultaneously feeding back the execution status to form a closed-loop control.

[0013] Preferably, in step S2, the standardized flexibility grading model uses the material's elastic modulus and deformation coefficient as core indicators, and weight and irregularity as correction indicators, to divide the package into 5 flexibility levels. The higher the level number, the higher the flexibility of the package.

[0014] Preferably, in step S3, a low-impact S-shaped segmented acceleration / deceleration command is generated and the unloading angle is reduced for highly flexible packages, a high-response segmented acceleration / deceleration command is generated and the unloading angle is increased for rigid packages, and a balanced control command is generated for semi-rigid or semi-flexible packages.

[0015] Preferably, the method further includes a sorting result closed-loop optimization and anomaly tolerance step: collecting package sorting results, optimizing corresponding control parameters based on misclassification and damage anomalies, and periodically optimizing the grading model and parameter mapping library; when anomalies such as recognition failure or location loss occur, triggering the fault tolerance mechanism to divert abnormal packages.

[0016] This technical solution has the following technical effects: 1. This invention takes the quantitative grading of package material flexibility as its core and establishes a linkage mapping relationship of "flexibility level - sorting failure mode - control parameters". It matches exclusive acceleration curves and unloading angles for packages with different flexibility, which fundamentally solves the problem of poor adaptability of fixed parameters in existing technologies. This helps to reduce the damage rate and misclassification rate of irregularly shaped packages while ensuring sorting efficiency and balancing sorting efficiency and package safety.

[0017] 2. This invention achieves non-destructive and rapid detection of the elastic modulus of packaging materials through hyperspectral material identification technology. Combined with 3D scanning and mechanical models, it realizes accurate quantification of the deformation coefficient, providing a reproducible and high-precision quantitative basis for flexibility grading, and avoiding the subjectivity and error of manual grading.

[0018] 3. This invention constructs an adaptive S-shaped acceleration and deceleration mathematical model and an unloading angle trajectory inversion model based on flexibility level, realizing accurate quantitative solution of control parameters instead of fixed lookup table matching. It can adapt to the full-scenario adaptive control of sorting lines of different specifications and different types of packages, and has strong generalization ability.

[0019] 4. The system architecture of this invention can be directly adapted to existing mainstream high-speed cross-belt sorting machines without large-scale modification of the main structure of the sorting line. The modification cost is low and it is easy to implement. It can be widely used in high-speed sorting scenarios of various irregular and flexible packages in e-commerce and express delivery industries, and has a wide range of applications. Attached Figure Description

[0020] Figure 1 This is a structural block diagram of the parcel sorting system described in an embodiment of the present invention.

[0021] Figure 2 This is a flowchart illustrating the control method of the parcel sorting system according to an embodiment of the present invention. Detailed Implementation

[0022] The embodiments of the present invention are described in detail below, and the present invention will be further explained in detail. This embodiment uses a high-speed cross-belt sorting machine commonly used in e-commerce express distribution centers with a linear speed of 2.5m / s, a trolley pitch of 300mm, and a sorting efficiency of 12,000 pieces / hour as a carrier to clearly and completely describe the technical solution of the present invention. The described embodiments are only a part of the embodiments of the present invention, and are not intended to limit the scope of protection of the present invention.

[0023] Example 1: Parcel Sorting System The package sorting system described in this embodiment has the following structure: Figure 1As shown, it includes a package information collection unit, a material flexibility level quantification and classification unit, a differentiated control instruction generation unit, a package ID binding and tracking unit, a microsecond-level control instruction scheduling center, a multi-axis servo execution unit, and a sorting result feedback and closed-loop optimization unit.

[0024] Package information collection unit: Deployed at the entrance loading section of the cross-belt sorting line, the weighing module, 3D laser contour scanning module, hyperspectral material recognition module, and high-speed vision module are arranged sequentially along the conveying direction. All modules are synchronously triggered by the same trigger signal output by the photoelectric sensor of the loading section, and the timestamp alignment error of the data collected by each module is ≤100μs.

[0025] The weighing module employs a high-precision dynamic weighing sensor with a range of 0.01-30kg and a weighing accuracy of ±2g, used to collect the package's weight (m) and center of gravity offset. The 3D laser contour scanning module uses a line laser contour sensor with a scanning frequency ≥1000Hz and a measurement accuracy of ±0.5mm, used to collect the package's three-dimensional dimensions and original height. Contact area between the package and the surface The system measures the cross-sectional area S of the package and identifies surface deformation features. The high-speed vision module uses a global shutter industrial camera with a frame rate of ≥200fps to capture the outer contour of the package, identify irregular package types, and calculate the irregularity of the package (irregularity = area of ​​the outer rectangle of the package / actual contour area of ​​the package).

[0026] Supplementary notes for the hyperspectral material identification module: The hyperspectral material identification module uses a pushbroom hyperspectral camera in the 400-1000nm band, with a spectral resolution ≤2nm and a spatial resolution ≤0.5mm. It is used to collect hyperspectral reflectance data of the packaging material and extracts the elastic modulus E of the material through a built-in material mechanical parameter inversion model. The specific implementation is as follows: (1) Hyperspectral data preprocessing: The original hyperspectral data were subjected to black and white correction, Savitzky-Golay smoothing and noise reduction, and multivariate scattering correction in sequence to eliminate ambient light interference, scattering errors caused by uneven surface of the package, and system noise, so as to obtain the standardized material spectral reflectance curve R(λ), where λ is the spectral wavelength.

[0027] (2) Feature band extraction: The continuous projection algorithm is used to extract the feature bands that are strongly correlated with the elastic modulus of the material from the full-band spectrum. The core feature bands include: 1210nm and 1730nm for polyethylene soft film, 1190nm and 1760nm for polypropylene woven bag, 1250nm and 1710nm for polyethylene terephthalate, 1450nm and 1930nm for corrugated paper or kraft paper, and 1680nm and 2130nm for hard plastic or metal. This eliminates redundant band interference and improves the inversion speed and accuracy.

[0028] (3) Elastic modulus inversion model: The model has a built-in pre-trained partial least squares regression material mechanical parameter inversion model. The model training set consists of 120 standard material samples, including 12 types of packaging materials commonly used in the express delivery industry. The standard elastic modulus of each sample was obtained by actual measurement using a universal testing machine. Simultaneously, corresponding hyperspectral data are collected; the model takes the spectral reflectance of the characteristic band as input and the material's elastic modulus E as output, with the model's coefficient of determination... With a value of ≥0.98 and a relative prediction error of ≤3%, it enables non-destructive, real-time detection of the elastic modulus of packaging materials.

[0029] (4) Multi-layer packaging scenario processing: For multi-layer packaging scenarios such as cardboard boxes covered with woven bags, the outermost and second outermost materials are identified by high-spectral depth resolution. The material with the smaller elastic modulus is taken as the grading benchmark. At the same time, the deformation of 3D scanning is used for cross-validation to ensure the accuracy of grading.

[0030] (5) Quantitative calculation of deformation coefficient: Combining the elastic modulus E with other collected parameters, the deformation coefficient ε of the package is calculated using the Hooke's Law mechanical model. The mathematical model is as follows: ε=

[0031] Where Δh is the static load deformation of the surface wrapped around the cross strip, and the calculation formula is: In the formula: m is the weight of the package, and g is the acceleration due to gravity (taken as 9.8 m / s²). Where E is the contact area between the package and the surface of the packing, E is the elastic modulus of the packaging material, and S is the load-bearing cross-sectional area of ​​the package. This represents the original height of the package.

[0032] Meanwhile, the 3D laser contour scanning module can directly measure the actual deformation of the package and cross-validate it with the calculated values ​​from the model to ensure the accuracy of the deformation coefficient.

[0033] (6) Dynamic load deformation correction: For dynamic load impact scenarios in high-speed sorting, a dynamic load correction coefficient k_d is introduced. The corrected dynamic load deformation coefficient ε_d=ε·k_d, where k_d is positively correlated with the sorting linear speed. In this embodiment, when the linear speed is 2.5m / s, k_d is taken as 1.2 to ensure that the grading model is adapted to the dynamic sorting conditions.

[0034] Material flexibility level quantification unit: Implemented using an embedded ARM processor, its input is connected to the output of the package information acquisition unit via gigabit Ethernet communication, and it incorporates a standardized package material flexibility grading model. This model is a grading decision tree built based on the C4.5 algorithm, using the material elastic modulus E and deformation coefficient ε as core grading indicators, and package weight m and irregularity as correction indicators, dividing packages into 5 flexibility levels. The specific grading rules are as follows: Level 1 (rigid): Elastic modulus E≥1000MPa, deformation coefficient ε≤0.5%, typical packaging is wooden box, metal box, hard plastic box; Level 2 (semi-rigid): Elastic modulus E=100-1000MPa, deformation coefficient ε=0.5%-2%, typical packaging is thickened corrugated cardboard box; Level 3 (semi-flexible): Elastic modulus E=10-100MPa, deformation coefficient ε=2%-5%, typical packages are ordinary corrugated cardboard boxes and express envelopes; Level 4 (Flexible): Elastic modulus E=1-10MPa, deformation coefficient ε=5%-15%, typical packaging is bubble wrap or thickened soft plastic bag; Level 5 (High Flexibility): Elastic modulus E≤1MPa, deformation coefficient ε≥15%, typical packaging includes woven bags, ultra-thin soft plastic bags, and soft-pack clothing.

[0035] After receiving the package feature parameters, the unit completes the level division through a hierarchical decision tree and outputs a unique corresponding flexibility level label k (k=1,2,3,4,5) for each package.

[0036] Differentiated control instruction generation unit: Implemented using an industrial controller, its input end is connected to the output end of the material flexibility level quantification unit, and it has a built-in preset flexibility level-sorting control parameter mapping library, an adaptive S-shaped acceleration and deceleration algorithm model based on flexibility level, and an unloading angle trajectory inversion model.

[0037] (1) Supplementary explanation of the adaptive S-shaped acceleration / deceleration algorithm model: This embodiment uses a 7-segment S-shaped acceleration and deceleration curve as the basic model of the acceleration curve. The curve is divided into acceleration segment, uniform acceleration segment, deceleration segment, uniform speed segment, acceleration and deceleration segment, uniform deceleration segment, and deceleration and deceleration segment, which can realize continuous change of acceleration and deceleration without step impact, perfectly adapting to the impact resistance requirements of different flexible wrapping.

[0038] Using the flexibility level k of the package as the core correction factor, an adaptive solution model for the core parameters of acceleration and deceleration is established, as follows: ① Adaptive correction formula for core parameters: Maximum jerk:

[0039] Maximum acceleration:

[0040] Total duration of acceleration phase:

[0041] In the formula: The reference maximum jerk for a level 1 rigid enclosure is taken in this embodiment. =8000 ; The reference maximum acceleration for a level 1 rigid enclosure is taken in this embodiment. =20 ; The total duration of the baseline acceleration segment for a level 1 rigid enclosure is taken in this embodiment. =80 ; α, β, and γ are the attenuation / growth coefficients calibrated in the experiment. In this embodiment, α = 0.4, β = 0.25, and γ = 0.3, respectively. k represents the level of flexibility in the wrapping material, where k = 1, 2, 3, 4, 5. The larger the k value, the more flexible the wrapping material becomes. , The smaller, The longer the acceleration and deceleration process, the smoother the acceleration and deceleration, and the smaller the impact.

[0042] ② Time function model of 7-segment S-shaped acceleration and deceleration: Let the acceleration phase duration be... The duration of the uniform acceleration phase is The duration of the deceleration and acceleration phase is The duration of the uniform speed segment is The duration of the acceleration / deceleration phase is The duration of the uniform deceleration segment is The deceleration phase duration is ,satisfy = = = Total duration T= + + + + + + .

[0043] jerk of each segment acceleration ,speed Displacement The time function is as follows (taking the acceleration phase as an example): Acceleration segment (0≤t< ): = , = , =, ; Uniform acceleration segment ( ≤t< + ): =0, = , = , ; deceleration phase ( + ≤t< + + ): = , - , = , In the formula, is the initial velocity of the acceleration phase. , , , This represents the velocity and displacement of the corresponding segment nodes.

[0044] The function model for the deceleration phase is symmetrical to that for the acceleration phase, and will not be elaborated upon here.

[0045] ③ Command Discretization: After obtaining the continuous acceleration curve based on the above model, it is discretized according to the control cycle of the servo driver (100μs in this embodiment) and converted into a position loop control command sequence in the CSP cyclic synchronous position mode. Each control cycle corresponds to a target position command, which is sent to the servo driver.

[0046] (2) Supplementary explanation of the unloading angle trajectory inversion model: The mechanical model of the package unloading and falling trajectory is established with the unloading point of the conveyor belt as the origin, the horizontal direction pointing towards the grid as the positive x-axis, and the vertical downward direction as the positive y-axis, forming a coordinate system. The falling motion of the package is simultaneously affected by the conveyor belt's driving force, friction, and gravity. The trajectory equations are as follows: Horizontal direction:

[0047] Vertical direction:

[0048] In the formula: This is the unloading angle (with surface rotation angle, counterclockwise is positive); The linear velocity of the surface at the start of unloading; The coefficient of dynamic friction between the wrapping and the surface is positively correlated with the flexibility level k. In this embodiment... ; It is the tangential acceleration due to surface rotation; H is the vertical height from the surface of the strip to the bottom plate of the grid. g is the acceleration due to gravity.

[0049] Solution rule: When the package is completely detached from the surface of the pack, The total time of the drop can be obtained by solving the problem. ,Will Substituting into the horizontal equation, we obtain the total drop distance X(θ,k). Let X(θ,k) = ( (where the horizontal distance is from the center of the target grid to the unloading origin) and the optimal unloading angle is obtained by inverse solving using Newton's iteration method. .

[0050] Newton's iteration convergence condition: The initial value of the iteration is the reference unloading angle. =30°, iteration step size 0.1°, convergence threshold is |X(θ,k)− |≤5mm, maximum number of iterations ≤20, when the maximum number of iterations is exceeded, the reference angle is taken to ensure the real-time performance and stability of the solution.

[0051] Generate corresponding unloading angle and position control commands, including parameters such as disc rotation angle, rotation direction, and holding time in position.

[0052] (3) Handling multiple packages in the same vehicle: For the scenario where the same vehicle carries two or more packages, control parameters are matched according to the package with the highest flexibility level, while limiting the maximum acceleration to ensure the sorting safety of all packages.

[0053] (4) Instruction generation process: After receiving the flexibility level label k of the package, the unit performs the following steps to generate control instructions: ① Enter the package's flexibility level k, weight m, and target compartment distance. Sorting line baseline parameters; ② Solving the core parameters of S-shaped acceleration and deceleration based on adaptive correction formula , , ; ③ Solve the continuous velocity / position curve based on the 7-segment S-shaped acceleration and deceleration model, and discretize it into a servo position loop control command sequence; ④ Based on the coefficient of friction Using the drop trajectory model, the optimal unloading angle is solved inversely. Generate unloading angle adjustment command; ⑤ Synchronize the acceleration curve command and the unloading angle command in time to ensure that the unloading action and the acceleration and deceleration process are accurately matched; ⑥ Output a complete set of control instructions, bind it to the global ID of the package, and send it to the microsecond-level control instruction scheduling center.

[0054] Package ID Binding and Tracking Unit: This unit includes an ID coding module and a real-time position tracking module, implemented using an industrial controller and distributed sensors. It is bidirectionally connected to the package information acquisition unit and the microsecond-level control command scheduling center. The ID coding module assigns a unique 32-bit global ID to each package entering the loading section, binding the package's characteristic parameters, flexibility level label, and control commands to the global ID. The real-time position tracking module deploys a set of photoelectric sensors and barcode readers every 10m along the sorting line conveyor direction, and an absolute encoder along the sorting line main shaft. With a sampling period ≤50μs, it collects the position and speed of the cross-belt trolleys and the package's on-time status in real time, updating the package global ID-trolley ID-real-time position mapping table in real time. The package position tracking error is ≤1mm.

[0055] The microsecond-level control command scheduling center is the core scheduling unit of this system. It adopts a hard real-time architecture using a Xilinx Zynq-7000 series FPGA and an EtherCAT industrial Ethernet bus. Its inputs are connected to the outputs of the differentiated control command generation unit and the package ID binding and tracking unit, respectively. Its outputs communicate with the servo motor drivers of the multi-axis servo execution unit via the EtherCAT bus. This scheduling center uses an EtherCAT distributed clock (DC) to achieve timing synchronization of all servo axes on the entire sorting line, with a synchronization error ≤1μs, a command scheduling cycle ≤100μs, and an end-to-end command issuance delay ≤50μs. It supports parallel scheduling of no fewer than 1000 servo axes.

[0056] Its core workflow is as follows: receive the control command corresponding to the global ID of the package, store it in the command cache queue, match it in real time with the package ID-cart ID-real-time location mapping table, pre-calculate the time when the package arrives at the target sorting grid trigger position, and within the scheduling cycle before the package arrives at the trigger position, accurately synchronize and send the control command of the package to the servo motor driver of the corresponding car to ensure that the execution of the command is completely synchronized with the position of the package.

[0057] Multi-axis servo actuators include servo motor drivers arrayed on each cross-belt trolley, permanent magnet synchronous servo motors, and reel sorting mechanisms. All servo motor drivers communicate with the microsecond-level control command scheduling center via EtherCAT bus. The servo motor drivers operate in CSP cyclic synchronous position mode, receiving control commands from the scheduling center to drive the servo motors and the reel mechanism to complete segmented accelerated transport, unloading angle adjustment, and precise unloading of packages. Simultaneously, the motor position, speed, torque, and command execution status are fed back to the microsecond-level control command scheduling center in real time with a period of ≤100μs, forming a closed-loop control at the execution layer.

[0058] The sorting result feedback and closed-loop optimization unit includes a high-speed visual inspection module and a through-beam photoelectric sensor deployed at the entrance of each sorting slot. Its outputs are connected to the inputs of the microsecond-level control command scheduling center and the differentiated control command generation unit, respectively. This unit collects sorting result data of packages, determines whether a package has fallen into the target slot, whether it is damaged, or whether it has slipped and been thrown away, and binds the sorting result to the package's global ID. If a package is misplaced, the unloading angle correction coefficient for the corresponding flexibility level is automatically optimized based on the misplacement offset. If damage or throwing occurs, the acceleration attenuation coefficient α and acceleration attenuation coefficient β for the corresponding flexibility level are automatically adjusted. The flexibility grading model, acceleration / deceleration algorithm parameters, and trajectory inversion model are periodically optimized based on full-volume sorting big data to ensure the sorting accuracy of the system during long-term operation.

[0059] Example 2: Control method for a parcel sorting system This embodiment is based on the system described in Embodiment 1, and the process is as follows: Figure 2 As shown, the specific steps include: S1 Package Feature Information Collection and ID Binding: When a package to be sorted enters the sorting line section, the photoelectric sensor in the upper section triggers all modules of the package information collection unit to work synchronously, collecting full quantitative feature information such as the material elastic modulus, deformation coefficient, packaging material type, weight, three-dimensional dimensions, center of gravity offset, and irregularity of the package. The timestamp alignment error of all collected data is ≤100μs. At the same time, the ID coding module assigns a unique 32-bit global ID to each package, binding the collected feature information to the global ID one by one.

[0060] S2 Package Material Flexibility Level Quantification: The collected package feature parameters are input into the preset standardized package material flexibility level decision tree model. The elastic modulus and deformation coefficient of the material are used as the core indicators, and the weight and irregularity are used as correction indicators to complete the quantitative classification of the package's flexibility level. A unique 1-5 level flexibility level label k is assigned to the package, and the level label is bound to the package's global ID.

[0061] S3 Differentiated Sorting Control Instruction Generation: Based on the preset adaptive S-shaped acceleration / deceleration algorithm model and unloading angle trajectory inversion model, and according to the package's flexibility level label k, the following steps are executed to generate control instructions: ① Solving for the core parameters of S-shaped acceleration and deceleration based on k , , Specifically: For Level 5 high-flexibility packages, ≤500r / The acceleration phase lasts for ≥200ms, and the unloading angle is reduced by 5°-10° compared to the reference angle. For level 3-4 semi-flexible / flexible wrapping, =500-2000r / The acceleration phase lasts 100-200ms, and the unloading angle matches the baseline angle. For level 1-2 rigid / semi-rigid enclosures, ≥2000r / The acceleration phase lasts ≤100ms, and the unloading angle is 3°-5° larger than the reference angle; ② Generate a segmented control command sequence for the acceleration curve based on a 7-segment S-shaped acceleration / deceleration model; ③ Based on the inverse solution of the drop trajectory model, the optimal unloading angle is obtained, and an unloading angle adjustment command is generated; ④ Synchronize the two sets of instructions in time and bind them to the global ID of the package before sending them to the instruction cache queue of the microsecond-level control instruction scheduling center.

[0062] S4 Package Real-time Tracking and Microsecond-Level Command Scheduling: Through the real-time position tracking module, the position, speed, and package status of the cross-belt trolley are collected in real time with a sampling period of ≤50μs, and the mapping table of package global ID-trolley ID-real-time position is updated, with a position tracking error of ≤1mm; The microsecond-level control command scheduling center realizes the timing synchronization of the servo axes of the entire system based on the EtherCAT distributed clock, with a synchronization error of ≤1μs and a scheduling cycle of ≤100μs; Based on the real-time position of the package and the layout of the sorting line grid, the scheduling center pre-calculates the time when the package arrives at the trigger position of the target sorting grid, and within one scheduling cycle before the package arrives at the trigger position, the control command of the package is accurately synchronized and sent to the servo motor driver of the corresponding trolley.

[0063] S5 sorting action execution and real-time feedback: The servo motor driver receives control commands issued by the scheduler, drives the servo motor and the tray mechanism, and strictly follows the commands to complete the segmented accelerated conveying, unloading angle adjustment and precise unloading and sorting of packages; at the same time, the position, speed, torque and command execution status of the motor are fed back to the microsecond-level control command scheduling center in real time with a period of ≤100μs, forming a closed-loop control at the execution layer.

[0064] S6 Sorting Result Closed-Loop Optimization and Anomaly Tolerance: The sorting result data of packages is collected by the detection module of the sorting grid to determine whether the package has been missorted, damaged, or thrown away, and the sorting result is bound to the global ID of the package; if an anomaly occurs, the control parameters of the corresponding flexibility level are automatically optimized based on the anomaly type and deviation amount; the flexibility grading model, acceleration / deceleration algorithm model and trajectory inversion model are optimized regularly based on the full sorting big data; if package recognition failure, location loss, instruction timeout, or servo execution anomaly occurs in the whole process, the fault tolerance mechanism is immediately triggered to divert the abnormal package to the abnormal grid, and the abnormal data is recorded at the same time to avoid sorting line blockage and ensure continuous system operation.

[0065] The above description is only a preferred embodiment of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A parcel sorting system, comprising a parcel information acquisition unit and a multi-axis servo execution unit, wherein the multi-axis servo execution unit includes an array of servo motor drivers and sorting execution mechanisms correspondingly connected to each driver, characterized in that, It also includes a material flexibility level quantification unit, a differentiated control instruction generation unit, a package ID binding and tracking unit, and a microsecond-level control instruction scheduling center; The input end of the material flexibility level quantification unit is connected to the output end of the package information collection unit. It has a built-in standardized package material flexibility grading model to receive package feature parameters, complete the material flexibility level quantification of the package to be sorted, and output the corresponding package flexibility level label. The differentiated control instruction generation unit is communicatively connected to the grade division unit and has a built-in flexibility level-sorting control parameter mapping library, which is used to match and generate corresponding package-specific acceleration curve segmentation control instructions and unloading angle adjustment instructions according to the received grade labels. The package ID binding and tracking unit communicates bidirectionally with the acquisition unit and the scheduling center, respectively. It is used to assign a unique global ID to each package, complete the binding of package information, control instructions and global ID, track the package location in real time and update the mapping table of ID and corresponding execution mechanism and real-time location. The microsecond-level control command scheduling center is communicatively connected to the command generation unit, the tracking unit, and each servo driver. It adopts a hard real-time architecture with a command scheduling cycle of ≤100μs and a multi-axis synchronization error of ≤1μs. It is used to receive control commands and, based on the mapping table, precisely synchronize and schedule the commands to the corresponding servo drivers within the microsecond-level time window when the package enters the target workstation, thereby driving the actuator to complete the precise sorting of the package.

2. The parcel sorting system according to claim 1, characterized in that, The package information acquisition unit includes a synchronously triggered 3D laser contour scanning module, a hyperspectral material recognition module, a weighing module, and a high-speed vision module. The timestamp alignment error of the data acquired by each module is ≤100μs. It is used to collect the elastic modulus, deformation coefficient, weight, external dimensions, and irregular features of the package material.

3. The parcel sorting system according to claim 1, characterized in that, The standardized flexibility grading model uses the material's elastic modulus and deformation coefficient as core grading indicators, and weight and irregularity as correction indicators. It divides the package into 5 flexibility levels through a grading decision tree. The higher the level number, the higher the package's flexibility and the weaker its impact resistance.

4. The parcel sorting system according to claim 1, characterized in that, The flexibility level-sorting control parameter mapping library matches corresponding control strategies based on the sorting failure modes of packages with different flexibility levels: high flexibility packages are matched with parameters for low-impact S-shaped segmented acceleration and deceleration and reduced unloading angle; rigid packages are matched with parameters for high-response segmented acceleration and deceleration and increased unloading angle; and semi-rigid / semi-flexible packages are matched with balanced control parameters.

5. The parcel sorting system according to claim 1, characterized in that, The microsecond-level control command scheduling center adopts a hard real-time architecture of FPGA + EtherCAT industrial Ethernet bus, and realizes full system timing synchronization based on EtherCAT distributed clock, with end-to-end command issuance delay ≤50μs.

6. The parcel sorting system according to claim 1, characterized in that, It also includes a sorting result feedback and closed-loop optimization unit, which includes a detection module deployed at the sorting grid to collect package sorting results, optimize corresponding control parameters based on missorted and damaged abnormal data, and optimize the grading model and parameter mapping library based on full sorting data.

7. The control method of the system according to any one of claims 1-6 comprises the following steps: S1: Collect full quantitative feature information of the packages to be sorted, assign a unique global ID to the package and complete the information binding; S2: Input the feature information into the preset standardized flexibility grading model, complete the quantitative division of the flexibility level of the wrapping material, generate the corresponding level label and bind it with the global ID; S3: Based on the preset flexibility level-sorting control parameter mapping library, generate package-specific acceleration curve segmentation control instructions and unloading angle adjustment instructions according to the level label. S4: Track the package location in real time and update the mapping table between the global ID and the corresponding sorting execution mechanism and real-time location. Through the microsecond-level control command scheduling center, the control commands are precisely and synchronously scheduled to the corresponding servo motor driver within the microsecond-level time window when the package enters the target sorting station. S5: The servo motor driver receives instructions and drives the sorting execution mechanism to complete the segmented accelerated conveying, unloading angle adjustment and precise sorting of packages, while feeding back the execution status to form a closed-loop control.

8. The control method according to claim 7, characterized in that, In step S2, the standardized flexibility grading model uses the material's elastic modulus and deformation coefficient as core indicators, and weight and irregularity as correction indicators, to divide the package into 5 flexibility levels. The higher the level number, the higher the flexibility of the package.

9. The control method according to claim 7, characterized in that, In step S3, a low-impact S-shaped segmented acceleration and deceleration command is generated for the highly flexible package and the unloading angle is reduced; a high-response segmented acceleration and deceleration command is generated for the rigid package and the unloading angle is increased; and a balanced control command is generated for the semi-rigid / semi-flexible package.

10. The control method according to claim 7, characterized in that, It also includes closed-loop optimization of sorting results and fault tolerance steps: collecting package sorting results, optimizing corresponding control parameters based on misclassification and damage anomalies, and regularly optimizing the grading model and parameter mapping library; when anomalies such as recognition failure or location loss occur, triggering the fault tolerance mechanism to divert abnormal packages.

Images