Repository management and control method based on digital twinborn model

Abstract

The invention discloses a repository management and control method based on a digital twin model, and relates to the field of repository operation and maintenance. The method comprises the following steps: creating a digital twin environment, and constructing a virtual repository of the equipment repository; synchronously updating the real-time operation data of the equipment repository to the virtual repository; when the warehouse-in and warehouse-out task is received, various implementation schemes of the warehouse-in and warehouse-out task are formulated; performing simulation deduction on each implementation scheme to obtain a time deduction result; an optimal implementation scheme is selected through operation efficiency evaluation; evaluating the health state of the equipment in the equipment repository by using the historical test data of the equipment; according to the health state of the equipment, determining equipment for transporting the equipment in and out of a warehouse; identifying the disturbance of the equipment repository in the process of executing the warehouse-in and warehouse-out task; an intelligent decision-making mechanism based on multi-agent reinforcement learning is adopted, a disturbance avoidance strategy is made in a self-adaptive mode, and execution of warehouse-in and warehouse-out tasks is adjusted. The management and control capability and the operation efficiency of the equipment repository can be improved.

Description

Technical Field

[0001] The present application relates to the field of storage warehouse operation and maintenance, and in particular to a storage warehouse management and control method based on a digital twin model. Background Art

[0002] With the extensive application of automation technology in the storage warehouse operation and maintenance scenarios, the storage warehouse operation and maintenance system based on automated equipment has gradually replaced traditional manual operations, improving the efficiency of equipment center warehouse operation and maintenance. However, there are still problems in the current storage warehouse operation and maintenance management and control, such as poor precision, low efficiency, insufficient dynamic response capabilities, and lack of foresight in decision-making, which makes it difficult to meet the needs of large-scale equipment rapid delivery under systematic combat conditions. Summary of the invention

[0003] The purpose of this application is to provide a storage depot management and control method based on a digital twin model, which can improve the management and control capabilities and operating efficiency of equipment storage depots.

[0004] To achieve the above objectives, this application provides the following solutions:

[0005] The present application provides a storage warehouse management and control method based on a digital twin model, comprising: creating a digital twin environment, and building a virtual storage warehouse of an equipment storage warehouse in the digital twin environment; collecting real-time operation data of the equipment storage warehouse, and synchronously updating it to the virtual storage warehouse; when receiving an entry and exit task, formulating multiple implementation plans of the entry and exit task; based on the real-time status of the virtual storage warehouse in the digital twin environment, simulating and deducing each implementation plan to obtain a time deduction result of each implementation plan; evaluating the operation performance of the time deduction result of each implementation plan, and selecting the optimal implementation plan; using equipment historical test data to evaluate the health status of equipment in the equipment storage warehouse; according to the health status of the equipment, determining the equipment for transporting equipment in and out of the warehouse in the optimal implementation plan, and performing the entry and exit task in the equipment storage warehouse according to the optimal implementation plan and the determined equipment; identifying the disturbance existing in the equipment storage warehouse when performing the entry and exit task; using an intelligent decision-making mechanism based on multi-agent reinforcement learning to adaptively formulate and adjust the disturbance avoidance strategy, and decompose the disturbance avoidance strategy into control instructions, which are sent to the equipment storage warehouse for implementation until the entry and exit task is completed.

[0006] According to the specific embodiments provided in this application, this application has the following technical effects:

[0007] The present application provides a storage depot management and control method based on a digital twin model, which closely couples digital twin technology with the operation and maintenance of equipment storage depots to form an overall solution for the operation process management of equipment storage depots covering monitoring, simulation, analysis, decision-making and regulation, thereby achieving the monitoring and control of all elements, all businesses and all processes in the operation of the equipment storage depot, and improving the management and control capabilities and operating efficiency of the equipment storage depot. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] In order to more clearly illustrate the technical solutions in the embodiments of the present application or related technologies, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0009] Figure 1 This is a schematic diagram of a process of a storage management and control method based on a digital twin model in one embodiment of the present application;

[0010] Figure 2 A schematic diagram of an equipment storage depot operation control mode driven by a digital twin provided in one embodiment of the present application;

[0011] Figure 3 A schematic diagram of constructing a data model provided in another embodiment of the present application;

[0012] Figure 4 A schematic diagram of the relationship between events, activities, and processes provided in another embodiment of the present application;

[0013] Figure 5 A schematic diagram of an event scheduling algorithm flow chart provided in another embodiment of the present application;

[0014] Figure 6 A schematic diagram of time task bottleneck degree data provided by another embodiment of the present application;

[0015] Figure 7 A schematic diagram of the overall architecture of a digital twin storage management and control solution provided in one embodiment of the present application;

[0016] Figure 8 A schematic diagram of a task planning interface provided by another embodiment of the present application;

[0017] Fig. 9 A schematic diagram of a simulation evaluation interface provided in another embodiment of the present application. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0019] In order to make the above-mentioned objects, features and advantages of the present application more obvious and easy to understand, the present application is further described in detail below with reference to the accompanying drawings and specific implementation methods.

[0020] The concept of digital twin can be traced back to 2003, when it was proposed by Professor Grieves in the Product Lifecycle Management (PLM) course at the University of Michigan. At that time, it was called the "mirror space model". After the concept was proposed, Grieves used the names of "information mirror model" and "digital twin" in his papers. Based on this, this application uses the concept of digital twin to integrate multiple emerging information technologies such as Internet of Things technology, simulation technology, modeling technology, data processing technology, machine learning technology, and human-computer interaction technology into the operation process control of the storage warehouse to improve the efficiency of storage warehouse control.

[0021] In an exemplary embodiment, Figure 1 As shown, a storage management method based on a digital twin model is provided, including the following steps 101 to 109. Among them:

[0022] Step 101: Create a digital twin environment, and build a virtual repository of the equipment repository in the digital twin environment.

[0023] Step 102: Collect the real-time operation data of the equipment storage warehouse and update it to the virtual storage warehouse synchronously.

[0024] Step 103: When receiving a storage in and storage out task, formulate multiple implementation plans for the storage in and storage out task.

[0025] Step 104: Based on the real-time status of the virtual storage in the digital twin environment, each implementation plan is simulated and deduced to obtain the time deduction result of each implementation plan.

[0026] Step 105: Evaluate the operational performance of the time simulation results of each implementation plan and select the optimal implementation plan.

[0027] Step 106: Use the equipment historical test data to evaluate the health status of the equipment in the equipment storage library.

[0028] Step 107: According to the health status of the equipment, determine the equipment for transporting the equipment in and out of the warehouse in the optimal implementation plan, and perform the in and out tasks in the equipment storage warehouse according to the optimal implementation plan and the determined equipment.

[0029] Step 108: Identify disturbances existing in the equipment storage depot when executing the in-and-out tasks.

[0030] Step 109: Adopt an intelligent decision-making mechanism based on multi-agent reinforcement learning to adaptively formulate and adjust the disturbance avoidance strategy, and decompose the disturbance avoidance strategy into control instructions, which are sent to the equipment storage depot for implementation until the storage and retrieval tasks are completed.

[0031] Implementing steps 101 to 109 above will tightly couple the digital twin technology with the operation and maintenance of the equipment depot, forming an overall solution for the equipment depot operation process management and control that covers monitoring, simulation, analysis, decision-making, and regulation, thereby achieving full monitoring and control of all elements, all businesses, and all processes during the operation of the equipment depot, and improving the management and control capabilities and operating efficiency of the equipment depot.

[0032] In order to meet the demand for digital twin technology in the management and control of equipment depots, a closed-loop management and control mode of equipment depots driven by digital twins is designed. The real-time data of the depot is synchronized in the twin space to realize the virtual-real fusion and state synchronization of the twin space for the depot. Based on the twin environment, during the task planning period, simulation and performance evaluation are carried out for different task types involved in the depot in the twin environment. The bottlenecks in the management and control of the depot are analyzed and the performance evolution laws are predicted through the twin data accumulated during the operation of the depot. Finally, a dynamic response mechanism is designed in the twin environment for the uncertain disturbance factors in the operation of the depot, and a scheduling strategy is formulated according to the real-time disturbance state through the multi-agent reinforcement learning method. The strategy is decomposed into control instructions and sent to the physical depot for implementation, so as to dynamically correct the operation and maintenance process of the depot. The operation and control mode of equipment depots driven by digital twins is as follows: Figure 2 shown.

[0033] In another exemplary embodiment of the present application, the digital twin environment for equipment storage mainly serves the management and control of the storage operation and maintenance process, giving the storage more accurate status monitoring, more reasonable task planning, faster storage operation disturbance response and more scientific decision-making strategy formulation, so as to achieve refined management and control for the storage. The physical storage is an objective entity collection in the storage, mainly including storage, transportation, measurement and design and other automation equipment, equipment, as well as people, auxiliary resources and environment. The physical workshop requires different types of internal resources to have the ability to perceive and integrate key state parameters. On the one hand, it ensures that the key parameters of the equipment can be connected through the TCP / IP interface protocol for collection; on the other hand, for equipment that cannot directly collect information, the perception of information is realized by adding sensors. The virtual repository is the physical model and behavior rules of the elements such as people, machines, objects, and environment in the repository. The virtual reality engine Unity3D software is used to render the elements in the virtual repository and model the interactive actions. The virtual presentation and human-computer interaction functions of the repository are realized through the particle effects, three-dimensional roaming, collision detection, scene switching and other functional modules in the Unity engine. The physical repository is modeled from the three levels of elements, behaviors, and rules. The established model is stored in the database, and the data is stored and read by connecting the database through a web page, forming a virtual portrait mapping of all elements of the repository based on web pages and three-dimensional plug-ins. Then the above step 101 can be replaced by the following steps 201 to 203:

[0034] Step 201: Model the virtual repository from three levels: elements, behaviors, and rules to obtain a virtual repository framework.

[0035] Step 202: Use the virtual reality engine Unity3D software to render the elements in the virtual repository framework, and implement interactive actions and virtual presentation to obtain the virtual repository.

[0036] Step 203: storing the virtual storage in a database, and storing and reading data by connecting the database via a web page, thereby forming a virtual portrait mapping of all elements of the storage based on the web page and the three-dimensional plug-in.

[0037] In another exemplary embodiment of the present application, based on the data collection of the IoT sensing elements in the field, the collected data needs to be cleaned and preprocessed. Due to the large number and variety of IoT sensing devices, the digital twin follows certain standards for data management and transmission. The basic focus of establishing a complete twin environment is the establishment of a data model. Through the integration and fusion of different twin model data, the model reflects the hierarchy and association between the data and expresses the process of the storage task operation. The data transmission text adopts the data synchronization construction and transmission mode of AML+OPCUA. AML follows an object-oriented information storage method, and can model the physical and logical components of the management and control platform as data objects. There is a hierarchical relationship between objects, and each object can contain multiple attributes such as logic, geometry, topology, and cloud top. Establish a data model for the physical storage from three perspectives: equipment, equipment, and vehicle. Then the storage form of the virtual storage in the above step 101 is:

[0038] PSF::=SR ni +ST nj +SP nk ;

[0039] SR i = {NU id ,NM id ,R type ,RG et ,RS et ,ACT et ,RD et ,SI et ,Info};

[0040] ST j = {NU jd ,NM jd ,T type ,TS et ,TP et ,TT et ,TC et ,Info};

[0041] SP k = {NU kd ,NM kd ,P type ,PL,PP,PC et ,PG et ,SI et ,Info};

[0042] Among them, PSF stands for equipment storage, SR ni Indicates a collection of resources in the library, ST nj Represents the set of task processes in the repository, SP nkRepresents storage equipment, + represents the relationship between resources, task processes and equipment, ::= represents the total geometric set of resources, tasks and equipment included in the storage; SR i Indicates the i-th resource in the resource collection, NU id Indicates the resource number, NM id Indicates the resource name, R type Indicates the resource type, RG et Represents a geometric set, RS et Represents a state set, ACT et Represents an action set, RD et Represents a dataset, SI et Indicates a data interface set, Info indicates other information; ST j represents the j-th task behavior in the task, NU jd Indicates the task number, NM jd Indicates the task name, T type Indicates the task type, TS et Indicates instruction set, TP et Represents the action set, TT et Represents the time requirement set, TC et Represents a constraint set; SP k represents the kth product in the equipment set, NU kd Indicates the equipment number, NM kd Indicates equipment model, P type Indicates equipment type, PL indicates equipment position, PP indicates equipment status, PC et Represents equipment history data set, PG et Represents a geometric set, SI et Represents a data interface set.

[0043] In another exemplary embodiment of the present application, by connecting with controllers such as existing storage equipment, conveying equipment, docking equipment, transfer equipment and test support equipment in the equipment storage warehouse, real-time data such as the equipment's operating speed, turning angle, temperature and humidity of the storage environment, equipment docking status, transfer action status, test evaluation results, etc. are uploaded and action control instructions are issued through the industrial Ethernet and the digital twin platform. Since the data collection methods of different underlying devices are diverse, the sources are wide, and the transmission protocols and interfaces are different, an OPC UA digital twin server is built. The data layer collects data from environmental sensors, AGV vehicle sensors, equipment status collectors, etc. through the UDP protocol. Based on real-time manufacturing data processing technology, the data is cleaned, classified, and added value. After establishing a data model and unifying the data description, the status data and real-time operation data are obtained and stored in the system real-time database. The data publishing service is used to submit it to the OPC UA client to realize the virtual layer with geometric models to characterize the physical storage warehouse, AGV behavior events, equipment storage status to express the dynamic process of the workshop, and logical models to express the workshop operation logic. In order to improve the real-time performance of data interaction, data nodes are merged and compressed, and the data transmitted by the physical storage is synchronized by creating a corresponding data update address space in the virtual storage, so as to achieve high-speed data flow between the virtual storage and the physical storage. Then the above step 102 can be replaced by the following steps 301 to 305:

[0044] Step 301: Build an OPC UA digital twin server.

[0045] Step 302: The OPC UA digital twin server collects real-time operation data of the equipment repository through the UDP protocol.

[0046] Step 303: pre-process the real-time operation data; the pre-processing includes cleaning, classification and value-added.

[0047] Step 304: Introduce the AML data exchange format, establish a unified data model for the pre-processed real-time operation data from four perspectives: instance level, role library, interface library and system unit library, and store it in the database.

[0048] Step 305: Use the data publishing service to synchronously update the data in the database to the virtual storage library.

[0049] In the current operating status monitoring, the twin platform displays equipment history information, equipment status information, inventory information, environmental information, etc. in real time, dynamically tracks the real-time status of each key networking element in the storage warehouse, and continuously updates it in the virtual storage warehouse.

[0050] The system unit library in step 304 includes storage rooms, storage areas, shelves and other levels, and is composed of reusable system component templates. The role library includes different interfaces and attributes. The interface library defines the basic interface and its rules. The instance level instantiates the role library by referencing the system unit library and the interface library. Finally, the internal element (IE) is used to model the physical components in the management and control platform, such as Figure 3 As shown. The corresponding OPC UA server is formed through the intermediate format XML, and the OPC UA client can directly read or subscribe to and listen to one or more node attributes stored in the server address space. Combining AML technology and OPC UA technology can effectively solve the problems of data integration and fusion between heterogeneous data. The data of different devices, different task processes, and equipment in different locations can be integrated in an orderly manner. The data model analyzes and combines the data in an orderly manner so that it can truly reflect the actual in-and-out task activities.

[0051] In another exemplary embodiment of the present application, event-driven simulation is used to describe the operation and maintenance task simulation. The simulation entities include system (entity set), model (abstract object), system state (variable set), entity object, attribute (entity feature description), state (activity in event-driven simulation), event (instantaneous operation or behavior), process (composed of several ordered events), etc. The operation relationship between events, activities and processes is as follows: Figure 4 shown.

[0052] A simulation clock is also required in the time-driven simulation process to indicate the change of simulation time. In a continuous simulation system, when the continuous model is discretized into a simulation model, the change of simulation time is based on the determination of the simulation step size, which can be a fixed step size or a variable step size.

[0053] The general process of event-driven simulation research includes system modeling, determining simulation algorithms, establishing simulation models, designing simulation programs, running simulation programs, and outputting and analyzing simulation results.

[0054] 1) System modeling. The simulation model of an event-driven system can generally be described in the form of a flow chart or a network diagram. They all reflect the process that temporary entities go through within the system, the effects of permanent entities on temporary entities, and the logical relationships between them.

[0055] 2) Select simulation algorithm. Event-driven simulation algorithms mainly include event scheduling method, activity scanning method and three-order scanning method.

[0056] 3) Establish a simulation model. Establish a computer model of the simulated system based on a known simulation algorithm. The establishment of a simulation model is actually the definition of the system window and the rules for event processing. Different simulation models need to be established according to different simulation algorithms.

[0057] 4) Design simulation program. The simulation program is the realization of the model, which can be written by yourself or programmed in simulation language.

[0058] 5) Analysis of simulation results. Since event-driven simulation systems are inherently random, the results obtained from each simulation run are only a sampling of random variables. The credibility of the simulation results and improving the confidence level of the simulation results are very important.

[0059] From the above process, it can be seen that the traditional event-driven simulation must first consider the characteristics of the simulation object, perform system modeling based on these characteristics, then select a targeted simulation algorithm, establish a simulation model of the simulation object based on the system model, and finally perform system program design. This simulation process tends to simulate specific simulation objects, has poor versatility, cannot achieve the reuse of simulation system programs and simulation models, and has a complex simulation design process. It uses event scheduling algorithms including event scheduling and activity scanning, and designs a general event scheduling strategy on this basis. The event scheduling algorithm process is as follows: Figure 5 See Figure 5 In the above step 104, an event-driven simulation is performed using an event scheduling algorithm when simulating and deducing each implementation scheme. The execution process of the event scheduling algorithm is as follows:

[0060] Initialize the simulation end time, events and the time when the events occur; set the simulation clock equal to the time when the events occur; determine the type of events that occur at the simulation clock moment; execute events of the event type; generate subsequent event types and occurrence times; if the simulation clock is less than or equal to the simulation end time, advance the simulation clock to the next earliest occurrence time and return to step "determine the type of events that occur at the simulation clock moment"; if the simulation clock is greater than the simulation end time, end.

[0061] Assume that the simulation clock is T and the event type is Ei (i = 1, 2, ..., n). The specific execution process of the event scheduling method is as follows:

[0062] Step 1: Initialization: set the simulation start time t0 and end time tend, set the entity initialization state, set the initial event and its occurrence time ts;

[0063] Step 2 Let T = ts;

[0064] Step 3 determines the event type Ei occurring at the current simulation clock time T, and performs subsequent event sorting operations according to the event type;

[0065] Step 4:

[0066]

[0067] Step 5 advances the simulation clock to the next earliest occurrence time and goes to Step 3;

[0068] Step 6 ends the simulation.

[0069] Through the above steps, the execution process of the event scheduling method is completed, among which Step 5 reflects the clock advancement mechanism of discrete event simulation. In continuous system simulation, the clock advancement is carried out according to a fixed time step, which is called the fixed time increment method. In discrete system simulation, the system state only changes at a series of random and discrete time points. This event advancement method is called the event increment method.

[0070] In another exemplary embodiment of the present application, the time deduction results in the above step 104 include the total time for completing the implementation plan, the first batch of delivery time and the average delivery time.

[0071] The general idea of ​​steps 103 to 105 is: when the warehouse receives a new task, based on the real-time status of the warehouse in the twin environment, models of the storage system, transportation system, equipment, and facilities in the warehouse are established, and according to the process arrangement results, the assumed time is automatically generated, and the OPM standard process editor is used to complete the editing of the task process, support the simulation of the task, and apply the general process deduction engine to the twin environment of the warehouse for the formulated task implementation plan. Combined with event assumptions and OPM to complete the simulation deduction, by deducing different plans, the total task completion time, the first batch of warehouse delivery time, the average warehouse delivery time and other deduction results are obtained for performance evaluation, so as to select the formulated task implementation plan.

[0072] In another exemplary embodiment of the present application, the above step 106 may be replaced by the following steps 401 to 402:

[0073] Step 401: Construct a health status assessment model using historical equipment test data.

[0074] Step 402: Determine the health status level and / or fault type of the equipment based on the current operation data of the equipment and using the health status assessment model.

[0075] In the historical data analysis, based on the collected and collated historical test data of equipment, including cleaning, preprocessing and feature engineering, vector machine (SVM), neural network (NeuralNetwork), decision tree (EdcisionTree) and other algorithms are used to divide the training set and test set for health status assessment model training and optimization, define evaluation indicators and fault type indicators to determine the equipment health status level and fault level, and then continuously optimize through model evaluation and verification. Use equipment historical test data to build a machine learning model to achieve equipment health status assessment, fault type classification and failure / degradation analysis.

[0076] In another exemplary embodiment of the present application, based on the clear evolution law of bottleneck problems and system performance, the task process is monitored by manufacturing IoT sensing environment, and the disturbance events in the generation process are detected by complex event processing methods, and the prototype of the bottleneck generation is traced. The task progress is quantified, and the gap between the actual and the plan is identified as a condition for intelligent decision-making, so as to evaluate the impact of the disturbance on the progress in real time as the judgment condition for triggering the intelligent decision-making mechanism. An intelligent decision-making mechanism based on multi-agent reinforcement learning is designed, and a holographic virtual environment of the cave is built as the learning environment of the agent. The closed-loop decision-making feedback control logic of the task process is realized through interaction in the physical cave. A multi-agent reinforcement learning method for intelligent decision-making is developed, and relevant state attributes, decision-making behaviors and reward functions are defined. Through the experience sharing mechanism E-MARL, collaborative decisions can be made for the same objective function, further triggering the dynamic response and adaptive decision-making modules in the digital twin environment. After monitoring and evaluating the disturbance phenomena existing in the current storage warehouse, multi-agent reinforcement learning is used to make decisions on path conflicts, emergency obstacle avoidance, etc. in the virtual operation environment of the storage warehouse. Based on the state perception, action selection and reward mechanism of the transportation equipment agent, inventory management agent and task scheduling agent, data such as storage status, route, and AGV operation status are obtained. The interactive learning between the agent and the virtual environment is realized by using value function estimation, policy gradient method, multi-agent learning algorithm, etc., and the agent continuously adjusts its own strategy to maximize the cumulative reward. The verified effective decision-making strategy is executed in the actual physical storage warehouse, thereby regulating the disturbance events in the storage warehouse. Then the above step 109 adopts an intelligent decision-making mechanism based on multi-agent reinforcement learning to adaptively formulate and adjust the disturbance avoidance strategy, which can be replaced by the following steps 501 to 504:

[0077] Step 501: Build a holographic virtual environment of the cave.

[0078] Step 502: Utilize multi-agent reinforcement learning to develop a disturbance avoidance strategy in the digital twin environment.

[0079] Step 503: The real-time operation data of the equipment storage warehouse in the in-and-out storage task is obtained through the intelligent agent, and interactive learning between the intelligent agent and the cave warehouse holographic virtual environment is carried out to continuously adjust the disturbance avoidance strategy.

[0080] Step 504: When the maximum cumulative reward is reached, an adaptively adjusted disturbance avoidance strategy is obtained.

[0081] In another exemplary embodiment of the present application, in the above step 107, during the execution of the in-and-out tasks in the equipment storage depot according to the optimal implementation scheme, the method can also quantify the task bottleneck degree of the equipment transporting equipment in and out of the depot at different times. In view of the bottleneck problems existing in the operation of the storage depot, such as the poor real-time performance of the control means, the opaque task management process, the information gap between the management layer and the execution layer, and the poor real-time control capability of the tasks, the decision-makers can clarify the location of the storage depot control bottleneck and its drift law through bottleneck analysis, and judge the performance evolution law of the future system based on the operation and maintenance data accumulated in the twin environment, so as to provide foresight for decision-making.

[0082] Based on the task analysis, the bottlenecks are first quantified and limited. The limitations include: tasks that change over time; equipment storage areas and locations are different in the warehouse; the process for each device (AGV vehicle) to complete the task will be different according to the storage location, and the execution will be different; each equipment may be planned into the task in the out-of-warehouse and in-warehouse tasks; the AGV vehicle's route is different according to the corresponding position of the equipment resources planned for the out-of-warehouse and in-warehouse tasks; equipment anomalies and obstacle avoidance during task execution; each out-of-warehouse and in-warehouse task requires testing and inspection; and the AGV vehicle needs to return to the parking position after completing the task.

[0083] Depending on the bottleneck problem encountered in the task, the task delay may be the delay left by the previous unit (called input delay) or the delay generated by the current unit when executing the task (called relative delay). Therefore, the output delay of a unit is determined by the input delay and its own relative delay, and the relative delay can reflect whether the output delay is alleviated or aggravated compared to the input delay. Therefore, based on the quantitative method, the total task delay is equal to the cumulative relative delay of the devices in the task, then:

[0084]

[0085] in, m represents the number of equipment ready for entry and exit, W represents the number of equipment, and N represents the number of completed task instructions. It represents the relative delay of the wth device to the completed task instruction n at time t. When there is no delay in entering or leaving the warehouse, Represents the sum of the relative delays of the wth device for the completed task instruction n at time t. Relative delay of the task It can be obtained by calculating the difference between the task progress and the plan.

[0086] Therefore, the degree of task bottleneck can be quantified by the ratio of the relative delay time of AGV at time t to the total delay time of all AGVs, which can be expressed as:

[0087]

[0088] in, Indicates the task bottleneck degree of the device at time t.

[0089] Therefore, at each time t in the task process, according to the ranking of the contribution ratio of different AGV relative delays to the total delay, the AGV task evaluation bottleneck with the highest relative delay to the total delay can be determined. As the task continues, the relative delay of AGVs under different tasks can be recorded, and different bottleneck levels can be obtained. Finally, the time bottleneck degree data is formed to quantify the task bottleneck degree of different AGVs at different times in the task, such as Figure 6 shown.

[0090] In another exemplary embodiment of the present application, after the above step 106, the method may further include: simulating, predicting and deducing the future evolution law of the equipment storage depot based on the operation and maintenance data accumulated in the digital twin environment.

[0091] In the overall architecture construction of the present application, based on the physical element composition and operation mode of a typical equipment storage depot, a data application and operation control mode of "historical data analysis - real-time status monitoring - future operation regulation" is designed in the digital twin space of the storage depot. The data collected at the physical layer of the storage depot is applied to the analysis of the storage depot operation process, and the control strategy is executed according to the real-time status of the storage depot reflected by the data, so as to feedback and correct the abnormal operation process of the storage depot, forming a closed-loop control mode of the twin space over the physical storage depot.

[0092] Based on the same inventive concept, the embodiment of the present application also provides a digital twin model-based storage control system for implementing the digital twin model-based storage control method involved above. The implementation solution provided by the device to solve the problem is similar to the implementation solution recorded in the above method, so the specific limitations in one or more digital twin model-based storage control system embodiments provided below can refer to the limitations of the digital twin model-based storage control method above, and will not be repeated here.

[0093] In an exemplary embodiment, the digital twin model-based storage management and control system is designed to meet the needs of various intelligent application upgrades, forming an overall solution for improving storage operation and maintenance efficiency that covers monitoring, simulation, analysis, decision-making, and regulation. The overall architecture of the digital twin storage management and control solution is as follows: Figure 7 See Figure 7 ,A storage warehouse management and control system based on the digital twin model includes: a virtual-reality mapping layer, a data transmission and processing layer, a twin service layer, and a twin demonstration and verification layer.

[0094] Virtual-reality mapping layer: The virtual-reality mapping layer consists of two parts: the physical storage library and the virtual storage library. The virtual end and the physical end synchronize states and interact with each other through data.

[0095] Data transmission layer: The data transmission layer is the functional level for realizing data communication, processing, integration, and storage of the digital twin workshop. The transmission processing layer is connected to the controllers of existing storage equipment, conveying equipment, docking equipment, transfer equipment, and test support equipment. Through industrial Ethernet and the digital twin platform, the real-time data such as the equipment's running speed, turning angle, temperature and humidity of the storage environment, equipment docking status, transfer action status, and test evaluation results are uploaded and action control instructions are issued.

[0096] Twin service layer: The twin service layer can also be called the analysis and decision-making layer. It is based on the virtual repository. In the virtual space, it analyzes the integrated historical data, monitors the operating status of the real-time collected data, and simulates, predicts and deduces the future evolution laws of the repository.

[0097] Demonstration and verification layer: mainly completes the construction of a digital twin storage warehouse management and control demonstration and verification platform, packages various functional modules into service packages, and develops core functions such as three-dimensional monitoring of equipment storage warehouses, operation and maintenance process status monitoring, equipment health management, dynamic logistics scheduling, and storage warehouse systematic simulation on the basis of basic functional modules. It provides an integrated management and control portal for storage warehouse operation and maintenance, and releases the platform in the form of an application. Users can remotely interact with the storage warehouse twin platform on a variety of user interaction terminals such as wearable VR glasses and tablet computers.

[0098] Figure 8 To create an interface for task planning, Fig. 9 It is the simulation evaluation interface.

[0099] In response to the demand for information and intelligent transformation of storage depot management, this application studies the storage depot management and control system based on the digital twin model (equipment storage depot management and control platform driven by digital twin). On the basis of sorting out the storage depot business needs and operation and maintenance processes, it proposes a multi-level twin storage depot architecture of "data collection layer-transmission processing layer-analysis and decision-making layer-twin service layer", and conducts "historical data analysis-real-time status monitoring-future operation and control" in the twin space to form a closed-loop control mode of the twin space for the physical storage depot. This application forms an overall solution for the storage depot operation process management and control covering monitoring, simulation, analysis, decision-making and control, which solves the monitorability and controllability of all elements, all businesses and all processes in the storage depot operation process, and lays the foundation for the further transformation and upgrading of the equipment storage depot management and control mode.

[0100] The advantages of this application are as follows:

[0101] Advantage 1: In response to the demand for information and intelligent transformation of warehouse management and control, we carried out research on the overall architecture of smart warehouse management and control driven by digital twins. Based on the analysis of warehouse business needs and operation and maintenance processes, we proposed a multi-level twin warehouse architecture of "data collection layer - transmission and processing layer - analysis and decision-making layer - twin service layer", forming an overall solution for warehouse operation process management and control covering monitoring, simulation, analysis, decision-making and regulation.

[0102] Advantage 2: A data application and operation control mode of "historical data analysis - real-time status monitoring - future operation regulation" is designed in the digital twin space of the storage warehouse. The data collected at the physical layer of the storage warehouse is applied to the analysis of the storage warehouse operation process, and the control strategy is executed according to the real-time status of the storage warehouse reflected by the data, so as to feedback and correct the abnormal operation process of the storage warehouse, forming a closed-loop control mode of the twin space for the physical storage warehouse.

[0103] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0104] This article uses specific examples to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method and core ideas of this application. At the same time, for those skilled in the art, according to the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting this application.

Claims

1. A storage management and control method based on a digital twin model, characterized in that: include: Create a digital twin environment and build a virtual repository of equipment repositories in the digital twin environment; Collect real-time operation data of equipment storage and update it to virtual storage synchronously; When receiving inbound and outbound tasks, formulate multiple implementation plans for the inbound and outbound tasks; Based on the real-time status of the virtual storage in the digital twin environment, each implementation plan is simulated and deduced to obtain the time deduction results of each implementation plan; Evaluate the operational performance of each implementation plan based on the time simulation results and select the best implementation plan; Use historical equipment test data to assess the health status of equipment in the equipment storage depot; Determine the equipment for transporting the equipment in and out of the warehouse in the optimal implementation plan according to the health status of the equipment, and perform the tasks of transporting the equipment in and out of the warehouse in the equipment storage warehouse according to the optimal implementation plan and the determined equipment; Identify disturbances in the equipment storage depot during the in-and-out missions; An intelligent decision-making mechanism based on multi-agent reinforcement learning is adopted to adaptively formulate and adjust the disturbance avoidance strategy, and the disturbance avoidance strategy is decomposed into control instructions, which are sent to the equipment storage depot for implementation until the warehousing and out-of-warehouse tasks are completed.

2. The storage management and control method based on the digital twin model according to claim 1 is characterized in that: Build a virtual repository of equipment storage in the digital twin environment, including: The virtual repository is modeled from three levels: elements, behaviors, and rules to obtain the framework of the virtual repository. The virtual reality engine Unity3D software is used to render the elements in the virtual repository framework, and to implement interactive actions and virtual presentation to obtain the virtual repository; The virtual storage is stored in a database, and data is stored and read by connecting the database via a web page, forming a virtual portrait mapping of all elements of the storage based on the web page and the three-dimensional plug-in.

3. The storage management and control method based on the digital twin model according to claim 1 is characterized in that: The storage form of the virtual repository is: PSF::=SR ni +ST nj +SP nk ; SR i ={NU id ,NM id ,R type ,RG et ,RS et ,ACT et ,RD et ,SI et ,Info}; ST j ={NU jd ,NM jd ,T type ,TS et ,TP et ,TT et ,TC et ,Info}; SP k ={NU kd ,NM kd ,P type ,PL,PP,PC et ,PG et ,ARE YOU et ,Info}; Among them, PSF stands for equipment storage, SR ni Indicates a collection of resources in the library, ST nj Represents the set of task processes in the repository, SP nk Represents storage equipment, + represents the relationship between resources, task processes and equipment, ::= represents the total geometric set of resources, tasks and equipment included in the storage; SR i Indicates the i-th resource in the resource collection, NU id Indicates the resource number, NM id Indicates the resource name, R type Indicates the resource type, RG et Represents a geometric set, RS et Represents the state set, ACT et Represents an action set, RD et Represents a dataset, SI et Indicates a data interface set, Info indicates other information; ST j represents the j-th task behavior in the task, NU jd Indicates the task number, NM jd Indicates the task name, T type Indicates the task type, TS et Indicates instruction set, TP et Represents the action set, TT et Represents the time requirement set, TC et Represents a constraint set; SP k represents the kth product in the equipment set, NU kd Indicates the equipment number, NM kd Indicates equipment model, P type Indicates equipment type, PL indicates equipment position, PP indicates equipment status, PC et Represents equipment history data set, PG et Represents a geometric set, SI et Represents a data interface set.

4. The storage warehouse management and control method based on the digital twin model according to claim 1 is characterized in that: Collect real-time operation data of the equipment storage and update it to the virtual storage synchronously, including: Build an OPC UA digital twin server; The OPC UA digital twin server collects real-time operating data from the equipment storage library through the UDP protocol; Preprocessing the real-time operation data; the preprocessing includes cleaning, classification and value-added; Introducing the AML data exchange format, establishing a unified data model for the pre-processed real-time operation data from four perspectives: instance level, role library, interface library, and system unit library, and storing it in the database; The data publishing service is used to synchronize the data in the database to the virtual repository.

5. The storage warehouse management and control method based on the digital twin model according to claim 1 is characterized in that: When simulating each implementation scheme, an event scheduling algorithm is used to perform event-driven simulation. The execution process of the event scheduling algorithm is as follows: Initialize simulation end time, events and event occurrence time; Let the simulation clock equal the time when the event occurs; determining the type of event that occurred at the simulation clock instant; Execute an event of the event type; Generate subsequent event types and occurrence times; If the simulation clock is less than or equal to the simulation end time, advance the simulation clock to the next earliest occurrence time, and return to step "determine the type of event that occurs at the simulation clock moment"; If the simulation clock is greater than the simulation end time, then the simulation ends.

6. The storage management and control method based on the digital twin model according to claim 1 is characterized in that: Use historical equipment test data to assess the health status of equipment in the equipment storage, including: Use historical equipment test data to build a health status assessment model; According to the current operation data of the equipment, the health status assessment model is used to determine the health status level and / or fault type of the equipment.

7. The storage management and control method based on the digital twin model according to claim 1 is characterized in that: Adopting an intelligent decision-making mechanism based on multi-agent reinforcement learning, adaptively formulate and adjust disturbance avoidance strategies, including: Build a holographic virtual environment of the cave; Develop disturbance avoidance strategies in digital twin environments using multi-agent reinforcement learning; The intelligent agent obtains the real-time operation data of the equipment storage warehouse during the in-and-out mission, and conducts interactive learning between the intelligent agent and the holographic virtual environment of the cave warehouse to continuously adjust the disturbance avoidance strategy; When the maximum cumulative reward is reached, an adaptively adjusted disturbance avoidance strategy is obtained.

8. The storage management and control method based on the digital twin model according to claim 1 is characterized in that: According to the optimal implementation plan, during the process of performing the in-and-out tasks in the equipment storage depot, the task bottleneck degree of the equipment transporting the equipment in and out of the depot at different times is quantified, specifically including: Using the formula Calculate the task bottleneck degree of the equipment transporting equipment in and out of the warehouse at each moment; among them, Indicates the task bottleneck degree of the device at time t, represents the sum of the relative delays of the wth device to the completed task instruction n at time t, represents the relative delay of the w-th device to the completed task instruction n at time t, W represents the number of devices, and N represents the number of completed task instructions.

9. The storage management and control method based on the digital twin model according to claim 1 is characterized in that: Use historical equipment test data to assess the health of equipment in the equipment storage, and then include: Based on the operation and maintenance data accumulated in the digital twin environment, the future evolution of the equipment storage depot is simulated, predicted and deduced.

10. The storage warehouse management and control method based on the digital twin model according to claim 1, characterized in that: The time simulation results include the total time to complete the implementation plan, the first batch of delivery time and the average delivery time.

Images