Four-way garage energy efficiency and wear balance scheduling method and system based on digital twinning

Abstract

The invention discloses a four-way garage energy efficiency and wear balance scheduling method and system based on digital twinning, and belongs to the technical field of intelligent scheduling, and the method comprises the steps: constructing a multi-target composite cost function of dynamic weighted fusion time cost, energy consumption cost and wear cost; according to the equipment operation parameters, predicting the residual service life and the health degree of the equipment through an equipment digital twin model; according to the multi-objective composite function, the remaining service life and the equipment health degree, generating an abrasion balance index and equipment scheduling cost fused with abrasion; sorting the equipment in the four-way garage according to the wear leveling index, and generating an equipment wear leveling scheduling strategy; and in combination with the equipment scheduling cost and the equipment wear balance scheduling strategy, carrying out space-time decomposition global dynamic task arrangement on the expected task to obtain a target scheduling scheme, and carrying out pre-deployment processing on the four-way vehicle in advance. According to the method, the fusion time, the energy consumption and the abrasion are optimized at the same time, and sustainable and efficient operation of the four-way garage system can be achieved.

Description

Technical Field

[0001] This application relates to the field of four-way vehicle scheduling technology, and in particular to a method and system for scheduling energy efficiency and wear balance of four-way parking garages based on digital twins. Background Technology

[0002] With the increasing demands for warehousing density and operational efficiency from e-commerce and manufacturing industries, heavy-duty four-way parking systems have become an important development direction for modern intelligent warehousing due to their high space utilization and flexibility. However, in actual operation, existing scheduling systems generally suffer from the following problems: High energy consumption and high electricity costs: In heavy-duty four-way parking garages, the large instantaneous current drawn during starting, acceleration, and lifting of four-way vehicles causes peak power consumption in the warehouse. This not only increases electricity expenses but may also require paying high demand charges, placing a significant economic burden on businesses.

[0003] Uneven equipment wear and tear and reduced lifespan: Traditional scheduling algorithms primarily pursue the shortest path or shortest time, resulting in frequent use of some vehicles and lanes while other equipment remains idle for extended periods. This uneven usage pattern leads to excessive wear on localized equipment, shortening the overall lifespan of the equipment and increasing maintenance and replacement costs.

[0004] Slow dynamic response and system congestion: When faced with sudden tasks or equipment failures, existing systems typically adopt a passive response approach, lacking proactive scheduling capabilities. This leads to delayed system response to emergencies, easily causing system-level congestion and affecting operational efficiency and the timeliness of order delivery. Summary of the Invention

[0005] The main objective of this application is to propose a method and system for energy efficiency and wear balancing scheduling of a four-way parking garage based on digital twins, aiming to achieve intelligent scheduling that considers time efficiency, energy consumption control and equipment wear balancing.

[0006] To achieve the above objectives, one aspect of this application proposes a four-way parking garage energy efficiency and wear balance scheduling method based on digital twins, the method comprising: Construct a multi-objective composite cost function that dynamically weights and integrates time cost, energy cost, and wear cost; Based on real-time collected equipment operating parameters, the remaining service life and health of the equipment are predicted through a digital twin model of the equipment; wherein, the twin in the digital twin model of the equipment includes a four-way vehicle, a hoist, and a tunnel track; Based on the multi-objective composite function, the remaining service life, and the equipment health, a wear leveling index and a device scheduling cost for fused wear are generated. The equipment in the four-way garage is sorted according to the wear leveling index of the equipment in the four-way garage, and an equipment wear leveling scheduling strategy is generated. Based on the order forecast results sent by the order forecast module of the warehouse management system, and combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, the expected tasks are decomposed into spatiotemporal global dynamic task orchestration to obtain the target scheduling scheme, and the four-way vehicles are pre-deployed in advance.

[0007] In some embodiments, constructing a multi-objective composite cost function that dynamically weights and fuses time cost, energy consumption cost, and wear cost includes the following steps: Determine the time cost based on the estimated task completion time and the average task completion time; Dynamically configure the time cost weighting coefficient based on the urgency of the task; The energy cost is determined based on the total energy consumption of the road section and the standard energy consumption value. The energy consumption cost weighting coefficient is dynamically adjusted according to the electricity price period; wherein, the standard energy consumption value is the measured value of the equipment completing one standard double cycle under rated load. The wear cost is determined based on the equivalent wear rate and standard wear rate of the equipment components in the four-way garage; The wear cost weighting coefficient is dynamically adjusted based on the equipment health status; wherein, the standard wear rate is calculated based on the cumulative equivalent hours from the time the new equipment is installed until its first major overhaul. A multi-objective composite cost function is constructed based on the time cost, the time cost urgency coefficient, the energy cost, the energy cost weighting coefficient, the wear cost, and the wear cost weighting coefficient.

[0008] In some embodiments, predicting the remaining lifespan and health of the equipment using a digital twin model based on real-time collected equipment operating parameters includes the following steps: Construct twin objects based on the four-way vehicle, hoist, and tunnel track; Based on the components of the twin object, a number of minimum replaceable units are determined; Collect equipment operation data for each of the minimum replaceable units; wherein, the equipment operation data includes load, operating intensity, and working time; The cumulative wear equation is calculated based on the equipment operating data to obtain the dimensionless cumulative wear value. A long short-term network model combining attention mechanism and linear regression is used to construct a device digital twin model based on the twin object; Using the digital twin model of the equipment, the remaining lifespan of the twin object is predicted based on the equipment's operating data. The prediction results are then weighted and averaged to obtain the remaining lifespan of the equipment. The equipment health status is calculated based on the dimensionless cumulative wear value and the expected working time.

[0009] In some embodiments, generating the wear leveling index and the equipment scheduling cost of fused wear based on the multi-objective composite function, the remaining useful life, and the equipment health includes the following steps: Based on the remaining service life and health status of the equipment, the equipment in the four-way garage is filtered to obtain usable equipment and usable lanes; For each available device, a wear leveling index is calculated based on the device health, task execution rate, and remaining service life of the available device; wherein, the target roadway is the available roadway matched with the available device; The wear and tear balance weighting coefficient is dynamically adjusted based on the actual electricity price. The equipment scheduling cost of fused wear is calculated based on the multi-objective composite function, the wear leveling index, and the wear leveling weight coefficient.

[0010] In some embodiments, the method further includes the following steps: For devices whose health status and remaining service life are both below a preset threshold, a maintenance work order is triggered.

[0011] In some embodiments, the step of sorting the equipment in the four-way parking garage according to the wear leveling index of the equipment in the four-way parking garage and generating an equipment wear leveling scheduling strategy includes the following steps: The equipment in the four-way parking garage is screened based on its health status and remaining service life to obtain qualified equipment; The qualified equipment is sorted in ascending order of wear leveling index to obtain a candidate equipment pool sequence; An improved bipartite graph algorithm is used for optimal task matching.

[0012] When the head area of ​​the same lane is occupied by multiple vehicles in the same time slice, the vehicle with the high wear leveling index is forced to decelerate, the equipment call cost is recalculated, until there is zero conflict, and the equipment wear leveling scheduling strategy is obtained.

[0013] In some embodiments, the step of performing spatiotemporal decomposition and global dynamic task orchestration on the expected tasks based on the order forecasting results sent by the order forecasting module of the warehouse management system, combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, to obtain the target scheduling scheme, includes the following steps: A spatial grid is constructed based on the warehouse top view, and a discrete time axis is constructed based on a preset minimum time slice; wherein, the spatial grid is used to determine the position coordinates of each piece of equipment in the four-way garage; the discrete time axis is used to determine the start and end times of the expected tasks; Based on the location coordinates, nodes, edges, and conflict matrix are determined, and then a lane conflict map is constructed. For each predicted storage location, a predicted task is generated based on the task priority and the order prediction result. The predicted tasks whose prediction probability reaches the expected probability are added to the quasi-task pool to obtain the quasi-task queue. The current pool of idle vehicles is determined based on the equipment wear leveling scheduling strategy. Based on the current idle vehicle pool and the quasi-task queue, with the goal of minimizing the device call cost, the spatiotemporal path is calculated, a pre-instruction queue is generated, and the four-way vehicles are pre-deployed in advance. When the actual task arrives, according to the principle of merging tasks in the same roadway, and based on the task execution order of tasks from farthest to nearest, the actual tasks are merged to obtain an optimized scheduling scheme. The optimized scheduling scheme and the pre-instruction queue are input into the desired lane conflict map. An improved tabu search algorithm is used to reorder the optimized scheduling scheme with the goal of minimizing the scheduling cost of the second device and achieving zero conflict, thereby generating a conflict-free task sequence and obtaining the target scheduling scheme.

[0014] In some embodiments, the method further includes the following steps: Configure the scheduled data return period; During the timed feedback cycle, if two four-way vehicles enter the same end area at the same time slot, the vehicle with lower health status or the non-emergency vehicle will slow down; if the conflict still occurs, the following vehicle will be controlled to make a brief temporary stop in the straight-ahead area.

[0015] To achieve the above objectives, another aspect of this application proposes a four-way parking garage energy efficiency and wear balancing scheduling system based on digital twins, the system comprising: The first module is used to construct a multi-objective composite cost function that dynamically weights and fuses time cost, energy cost, and wear cost; The second module is used to predict the remaining service life and health of the equipment based on the real-time collected equipment operating parameters through a digital twin model of the equipment; wherein, the twin in the digital twin model of the equipment includes a four-way vehicle, a hoist, and a tunnel track; The third module is used to generate a wear leveling index and a device scheduling cost for merging wear based on the multi-objective composite function, the remaining service life, and the device health. The fourth module is used to sort the equipment in the four-way garage according to the wear leveling index of the equipment in the four-way garage and generate an equipment wear leveling scheduling strategy. The fifth module is used to perform spatiotemporal decomposition of the expected tasks and global dynamic task orchestration based on the order prediction results sent by the order prediction module of the warehouse management system, combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, to obtain the target scheduling scheme, and to pre-deploy the four-way vehicles in advance.

[0016] To achieve the above objectives, another aspect of this application provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the method described above.

[0017] To achieve the above objectives, another aspect of the embodiments of this application proposes a computer-readable storage medium storing a computer program that, when executed by a processor, implements the aforementioned method.

[0018] The embodiments of this application include at least the following beneficial effects: This application provides a method and system for energy efficiency and wear balancing scheduling of four-way parking garages based on digital twins. This scheme constructs a multi-objective composite cost function that dynamically weights and integrates time cost, energy consumption cost, and wear cost, thus combining the three costs of time, energy consumption, and wear. It achieves simultaneous optimization of efficiency, energy consumption, and lifespan in the field of heavy-duty four-way parking garages. Furthermore, it uses digital twins to predict the remaining lifespan of equipment and allocates tasks according to the remaining lifespan of the equipment to achieve wear balancing of all equipment in the garage. In addition, the embodiments of this application also pre-deploy four-way vehicles in advance by using spatiotemporal decomposition of global dynamic task orchestration, combined with equipment scheduling cost and equipment wear balancing scheduling strategy, thereby improving the system operating efficiency. Attached Figure Description

[0019] The accompanying drawings are used to provide a further understanding of the technical solutions of this application and constitute a part of the specification. They are used together with the embodiments of this application to explain the technical solutions of this application and do not constitute a limitation on the technical solutions of this application.

[0020] Figure 1 This is a flowchart of the method for scheduling energy efficiency and wear balance of a four-way parking garage based on digital twins, provided in an embodiment of this application. Figure 2 This is a flowchart provided in the embodiments of this application; Figure 3 This is a data processing flowchart based on a multi-objective composite cost function provided in an embodiment of this application; Figure 4 This is a data processing flowchart for predicting equipment health and generating wear leveling strategies based on a digital twin model of the equipment, provided in an embodiment of this application. Figure 5 This is a flowchart of the device health prediction process provided in the embodiments of this application; Figure 6 This is a flowchart illustrating the wear leveling strategy scheduling process provided in this application embodiment. Figure 7This is a data processing flowchart for global dynamic task orchestration based on spatiotemporal decomposition provided in an embodiment of this application; Figure 8 This is a schematic diagram of the modules of the four-way garage energy efficiency and wear balance scheduling system based on digital twin provided in the embodiments of this application; Figure 9 This is a schematic diagram of the hardware structure of the electronic device provided in the embodiments of this application. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of this application and are not intended to limit it. In the following description, when referring to the accompanying drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with those of this application; they are merely examples of apparatuses and methods consistent with some aspects of the embodiments of this application as detailed in the appended claims.

[0022] Although functional modules are divided in the system diagram and a logical order is shown in the flowchart, in some cases, the steps shown or described may be performed in a different order than the module division in the system or the order in the flowchart. The terms "first / S100," "second / S200," etc., in the specification, claims, and the aforementioned figures are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0023] It is understood that the terms “first,” “second,” etc., used in this application may be used herein to describe various concepts, but unless otherwise stated, these concepts are not limited by these terms. These terms are only used to distinguish one concept from another. For example, without departing from the scope of the embodiments of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the words “if,” “when,” or “in response to a determination” as used herein may be interpreted as “when…” or “when…” or “in response to a determination.”

[0024] As used in this application, the terms "at least one", "multiple", "each", "any", etc., "at least one" includes one, two or more, "multiple" includes two or more, "each" refers to each of the corresponding multiples, and "any" refers to any one of the multiples.

[0025] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0027] This application can be used in a wide variety of general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronics, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices. This application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc., that perform specific tasks or implement specific abstract data types. This application can also be practiced in distributed computing environments where tasks are performed by remote processing devices connected via a communication network. In distributed computing environments, program modules can reside in local and remote computer storage media, including storage devices.

[0028] Before providing a detailed description of the embodiments of this application, some of the nouns and terms involved in the embodiments of this application will be explained first. The nouns and terms involved in the embodiments of this application are subject to the following interpretations.

[0029] HI: Health indicators of the equipment.

[0030] RUL: Percentage of remaining useful life of the equipment.

[0031] MBI: Wear Leveling Index.

[0032] SLA: Expected completion time of the mission.

[0033] In related technologies, with the increasing demands for warehousing density and operational efficiency from e-commerce and manufacturing industries, heavy-duty four-way parking systems have become an important development direction for modern intelligent warehousing due to their high space utilization and high flexibility. However, in actual operation, existing scheduling systems generally suffer from the following problems: High energy consumption and high electricity costs: In heavy-duty four-way parking garages, the large instantaneous current drawn during starting, acceleration, and lifting of four-way vehicles causes peak power consumption in the warehouse. This not only increases electricity expenses but may also require paying high demand charges, placing a significant economic burden on businesses.

[0034] Uneven equipment wear and tear and reduced lifespan: Traditional scheduling algorithms primarily pursue the shortest path or shortest time, resulting in frequent use of some vehicles and lanes while other equipment remains idle for extended periods. This uneven usage pattern leads to excessive wear on localized equipment, shortening the overall lifespan of the equipment and increasing maintenance and replacement costs.

[0035] Slow dynamic response and system congestion: When faced with sudden tasks or equipment failures, existing systems typically adopt a passive response approach, lacking proactive scheduling capabilities. This leads to delayed system response to emergencies, easily causing system-level congestion and affecting operational efficiency and the timeliness of order delivery.

[0036] In view of this, this application provides a method and system for energy efficiency and wear leveling scheduling of a four-way parking garage based on digital twins. This scheme constructs a multi-objective composite cost function that dynamically weights and integrates time cost, energy consumption cost, and wear cost. Based on real-time collected equipment operating parameters, it predicts the remaining service life and health of the equipment using a digital twin model. The twins in the equipment digital twin model include the four-way vehicle, the hoist, and the tunnel track. Based on the multi-objective composite function, remaining service life, and equipment health, a wear leveling index and a device scheduling cost integrating wear are generated. The equipment in the four-way parking garage is sorted according to the wear leveling index, generating a wear leveling scheduling strategy. Based on the order prediction results sent by the order prediction module of the warehouse management system, combined with the device scheduling cost and the wear leveling scheduling strategy, the expected tasks are spatiotemporally decomposed into global dynamic task orchestration to obtain the target scheduling scheme.

[0037] Figure 1 This is an optional flowchart of the four-way parking garage energy efficiency and wear balance scheduling method based on digital twins provided in the embodiments of this application. Figure 1 The method may include, but is not limited to, the following steps S100 to S500.

[0038] Step S100: Construct a multi-objective composite cost function that dynamically weights and integrates time cost, energy cost, and wear cost.

[0039] Specifically, the expression for the multi-objective composite cost function constructed in this application embodiment is as follows: Total cost = α × time cost + β × energy cost + γ × wear and tear cost; Among them, α, β, and γ are dynamic weighting coefficients. These weighting coefficients are adjusted in real time according to the urgency of the task (order priority, delivery time limit), real-time electricity price band (such as peak-valley flat electricity price), and equipment health. Equipment health is predicted through the equipment digital twin model.

[0040] Furthermore, α is the time cost weighting coefficient, driven by task urgency, used to measure the penalty intensity for "late task delivery." Urgent orders require a larger α to prioritize timeliness. The dynamic adjustment of α is based on the following: configuring the task urgency factor f_u: f_u = 1 - (remaining time until SLA deadline) / (maximum allowed response time). The less time remaining, the closer f_u is to 1, and the larger α becomes; configuring the system congestion factor f_c: when the task queue depth > threshold, f_c slightly increases α to prevent systemic delays.

[0041] For example: when processing urgent orders (outbound within 30 minutes), if f_u = 0.15, then α will automatically rise to 0.5~0.6; when processing regular orders (outbound within 4 hours), if f_u = 0.2, then α will drop to 0.1~0.2.

[0042] β is the energy consumption cost weighting coefficient, driven by electricity price fluctuations, used to measure the intensity of the penalty for "electricity cost". During peak electricity price periods, β needs to be increased to guide vehicles to slow down and postpone tasks. The dynamic adjustment of β is based on the following: configuring the electricity price factor f_p: f_p=0~1, where 0 represents off-peak electricity and 1 represents peak electricity, using linear interpolation; configuring the demand approximation factor f_d: when the real-time power is close to 90% of the monthly maximum demand, f_d further increases β.

[0043] For example, during off-peak hours (23:00-07:00), f_p=0, and β is adjusted to ≈0.05~0.1 (energy consumption is almost ignored); during normal hours, fp=0.5, and β is adjusted to ≈0.2~0.3; during peak hours (10:00~12~00), fp=1, and β is adjusted to ≈0.5~0.6 (energy consumption is the main factor); during demand warning, f_d is superimposed, and β is temporarily adjusted to >0.7 to force a speed reduction and peak shaving.

[0044] γ is the wear cost weighting coefficient, driven by equipment health, used to measure the degree of penalty for "equipment lifespan depletion". Equipment with poor health requires a larger γ to avoid overuse. The dynamic adjustment of γ is based on the following: configuring the health factor f_h: f_h = 1 – (system average HI) / 100. The lower the average health, the closer f_h is to 1, and the larger γ is; configuring the single equipment over-limit factor f_e: if a certain equipment's RUL < 72h, f_e locally amplifies the γ of that equipment by 2 times, accelerating the retirement of that equipment.

[0045] For example, when all equipment in the warehouse is in good health (average HI=85%): f_h=0.15, adjust γ≈0.15 (lifespan secondary); when the health of all equipment in the warehouse deteriorates (average HI=40%): f_h=0.6, adjust γ≈0.35 (lifespan priority); if a single device's RUL<72 h, adjust f_e=2.0, adjust the device's γ_local=0.7, and the device will be automatically blocked by the scheduling system.

[0046] For the dynamic weighting coefficients α, β, and γ, Softmax normalization is used to ensure that α + β + γ = 1 and that the dynamic response is sensitive. The expression for Softmax normalization is as follows: exp_u = exp(k_u·f_u), exp_p = exp(k_p·f_p), exp_h = exp(k_h·f_h); α= exp_u / (exp_u + exp_p + exp_h); β= exp_p / (exp_u + exp_p + exp_h); γ= exp_h / (exp_u + exp_p + exp_h); Among them, k_u, k_p, and k_h are all sensitivity coefficients, and their values ​​can be corrected through online learning.

[0047] The adjustment cycles for α, β, and γ can be set to refresh once every 30 seconds, or other suitable adjustment cycles can be set to ensure a rapid response to changes in electricity prices and health status. This application does not impose any restrictions on this.

[0048] In general, the higher the electricity price, the larger the β, and the more energy-efficient the system; the more urgent the order, the larger the α, and the faster the system works; the older the equipment, the larger the γ, and the longer the system's lifespan.

[0049] Furthermore, in the multi-objective composite function, the time cost is determined based on the estimated task completion time and the average task completion time; the energy cost is determined based on the total energy consumption of the road segment and the standard energy consumption value; wherein, the standard energy consumption value is the measured value of the equipment making one standard double cycle under rated load; and the wear cost is determined based on the equivalent wear rate and standard wear rate of the equipment components in the four-way garage.

[0050] In order to transform the time cost, energy consumption cost, and wear cost into additive values ​​of the same dimension, this application establishes an original quantitative model for the time cost, energy consumption cost, and wear cost, as follows: For time cost, time cost T_cost = (estimated route travel time + loading and unloading time + queuing waiting time) / standard time unit; the standard time unit is the historical average single task cycle (e.g., 120s), ensuring that T_cost is dimensionless and the central value ≈ 1.

[0051] For energy consumption cost, E_cost = total energy consumption of the road segment / standard energy consumption, where total energy consumption of the road segment = peak starting current + energy consumption during constant speed phase + energy consumption during acceleration phase; standard energy consumption is the average value measured under rated load for a "typical round trip" (e.g., 0.25 kWh). Furthermore, the peak starting current is calculated using the motor torque-speed model integral; the energy consumption during acceleration phase is calculated using the direct formula m·g·Δhz; if energy feedback is included during deceleration phase, it is reduced according to the feedback efficiency.

[0052] The wear cost is calculated as W_cost = Σ(component wear factor × usage intensity) / standard wear amount. Components include the traveling wheels, lifting chain, motor bearings, and track. The usage intensity for these components corresponds to the number of start-stop cycles, the square integral of acceleration, lifting height, and load. The remaining lifespan percentage (RUL) of each component is calculated using a digital twin model of the equipment. After normalization, 1 - RUL represents the current wear factor. The standard wear amount is taken as 1 / 1000 of the cumulative wear value from new equipment to the first major overhaul, ensuring that the center value of W_cost ≈ 1. In some embodiments, other suitable ratios can be used to calculate the cumulative wear value.

[0053] Through the above step S100, a multi-objective composite cost function is constructed, which dynamically weights and integrates time cost, energy consumption cost and wear cost. The weight coefficients α, β and γ are adjusted in real time according to the real-time electricity price, equipment health and order urgency, so as to achieve peak shaving and valley filling and wear balance optimization.

[0054] Step S200: Based on the real-time collected equipment operating parameters, predict the remaining service life and health of the equipment through the equipment digital twin model; wherein, the twin in the equipment digital twin model includes the four-way vehicle, the hoist, and the tunnel track.

[0055] Specifically, step S200 includes the following steps: A digital twin object is constructed based on the four-way vehicle, hoist, and tunnel track. Several minimum replaceable units are determined according to the components of the digital twin object. Equipment operation data for each minimum replaceable unit is collected, including load, working intensity, and working time. A cumulative wear equation is calculated based on the equipment operation data to obtain the dimensionless cumulative wear value. A long-short term network model combining attention mechanism and linear regression is used to construct a digital twin model of the equipment based on the digital twin object. The remaining service life of the digital twin object is predicted based on the equipment operation data using the digital twin model. The prediction results are then weighted and averaged to obtain the remaining service life of the equipment. The equipment health is calculated based on the dimensionless cumulative wear value and the expected working time.

[0056] Step S300: Based on the multi-objective composite function, remaining service life, and equipment health, generate the wear leveling index and the equipment scheduling cost of fused wear.

[0057] Specifically, step S300 includes the following steps: By combining the remaining service life and health of the equipment, the equipment in the four-way garage is filtered to obtain available equipment and available lanes. For each available piece of equipment, a wear leveling index is calculated based on the equipment health, task execution rate, and remaining service life. The target lane is the available lane matched with the available equipment. The wear leveling weight coefficient is dynamically adjusted based on the actual electricity price. The equipment scheduling cost of merging wear is calculated based on the multi-objective composite function, the wear leveling index, and the wear leveling weight coefficient.

[0058] In some embodiments, a maintenance work order is triggered for devices whose health status and remaining service life are both below a preset threshold.

[0059] Step S400: Sort the equipment in the four-way garage according to the wear balance index of the equipment in the four-way garage, and generate the equipment wear balance scheduling strategy.

[0060] Specifically, step S400 includes: screening equipment in the four-way garage based on equipment health and remaining service life to obtain qualified equipment; sorting the qualified equipment in ascending order of wear leveling index to obtain a candidate equipment pool sequence; and using an improved bipartite graph algorithm for optimal task matching. When the head area of ​​the same lane is occupied by multiple vehicles in the same time slice, vehicles with high wear leveling indices are forced to decelerate, and the equipment call cost is recalculated until zero conflict is reached, thus obtaining the equipment wear leveling scheduling strategy.

[0061] Through the above steps S200~S400, the system can complete the closed loop of "collection-inference-balancing-scheduling" in a short time, extend the track life and the overhaul cycle of the four-way vehicle, and the health status does not need to be manually determined throughout the process.

[0062] Step S500: Based on the order forecast results sent by the order forecast module of the warehouse management system, and combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, the expected tasks are decomposed into spatiotemporal global dynamic task orchestration to obtain the target scheduling scheme, and the four-way vehicles are pre-deployed in advance.

[0063] Specifically, this mainly involves combining the WMS order forecasting module to predict tasks that may arise in the next few minutes and pre-arrange idle vehicles to wait in the "reserve area". It analyzes the pick-up and drop-off locations of all current tasks, intelligently merges multiple tasks in the same lane, and adjusts the task order to avoid "advance-reverse conflicts" or "reversing congestion" at the same lane end.

[0064] Further, step S500 includes the following steps: A spatial grid is constructed based on the warehouse top view, and a discrete time axis is constructed based on the preset minimum time slice. The spatial grid is used to determine the position coordinates of each piece of equipment in the four-way garage. The discrete time axis is used to determine the start and end times of the expected tasks. Nodes, edges, and conflict matrices are determined based on the position coordinates, and then a lane conflict map is constructed. For each predicted storage location, expected tasks are generated based on task priority and order prediction results. Expected tasks with predicted probabilities reaching the expected probabilities are added to a quasi-task pool to obtain a quasi-task queue. The current idle vehicle pool is determined based on the equipment wear leveling scheduling strategy. Based on the current idle vehicle pool and the quasi-task queue, the spatiotemporal path is calculated with the goal of minimizing equipment call cost, a pre-instruction queue is generated, and four-way vehicles are pre-deployed in advance. When actual tasks arrive, according to the principle of merging within the same lane, the actual tasks are merged based on the task execution order from farthest to nearest, resulting in an optimized scheduling scheme. The optimized scheduling scheme and the pre-instruction queue are input into the lane conflict map, and an improved tabu search algorithm is used to reorder the optimized scheduling scheme with the goal of minimizing the second equipment scheduling cost and zero conflict, generating a conflict-free task sequence to obtain the target scheduling scheme.

[0065] In some embodiments, step S500 further includes the following steps: configuring a timed feedback cycle; during the feedback cycle, when two four-way vehicles enter the same end area in the same time slot, the vehicle with lower device health or the non-emergency vehicle slows down; if the conflict still occurs, the following vehicle is controlled to make a brief temporary stop in the straight-ahead area.

[0066] Based on this, the embodiments of this application have at least the following beneficial effects: A multi-objective composite cost function is constructed, which dynamically weights and integrates time, energy, and wear costs. This function optimizes efficiency, energy consumption, and lifespan simultaneously in the heavy-duty four-way parking garage field. Furthermore, digital twins are used to predict the remaining lifespan of equipment, and tasks are assigned based on the remaining lifespan to achieve wear balance across all equipment in the garage. In addition, this embodiment of the application also pre-deploys four-way vehicles in advance by using spatiotemporal decomposition of global dynamic task orchestration, combined with equipment scheduling costs and equipment wear balance scheduling strategies, thereby improving system operating efficiency.

[0067] The following section provides a detailed description and explanation of the solutions in the embodiments of this application, using practical application examples: This application provides a digital twin-based method for scheduling energy efficiency and wear balancing in four-way parking garages. This method can be applied to scheduling energy efficiency and wear balancing in four-way parking garages. Specifically, the method of this application is described in detail with reference to practical applications, such as... Figure 2 As shown, during scheduling, steps 1 to 4 can be executed. Unless otherwise specified, the specific values ​​of the parameters mentioned below are for illustrative purposes only, and this application does not limit the specific values ​​of these parameters: Step 1: Construct a multi-objective composite cost function: Total cost = α × Time + β × Energy consumption + γ × Wear rate; In practical applications, such as... Figure 3 As shown, step 1 may include the following steps 1.1 to 1.7: Step 1.1: Establish three types of original cost quantification models.

[0068] In order to transform time, energy consumption, and wear into additive "numerical values ​​of the same dimension", a primitive cost quantification model for time cost, energy consumption cost, and wear cost is established.

[0069] Wherein, the time cost T_cost = (estimated route travel time + loading and unloading time + queuing waiting time) / standard time unit; the standard time unit is the historical average single task cycle (e.g., 120s), ensuring that T_cost is dimensionless and the central value ≈ 1.

[0070] Energy cost E_cost = Total energy consumption of the road segment (starting current peak + constant speed stage energy consumption + lifting stage energy consumption) / standard energy consumption; standard energy consumption is the average value measured under rated load for a "typical round trip" (e.g., 0.25 kWh). Furthermore, the starting current peak is calculated using the motor torque-speed model integral; the lifting stage energy consumption is calculated directly using the formula m·g·Δhz; if the deceleration section has energy feedback, it is reduced according to the feedback efficiency.

[0071] Wear cost W_cost = Σ(component wear factor × usage intensity) / standard wear amount, where components include wheels, lifting chains, motor bearings, and tracks, and the corresponding usage intensity includes start-stop frequency, integral of the square of acceleration, lifting height, and load. The digital twin model of the equipment calculates the remaining life percentage (RUL) of each component, and after normalization, 1-RUL is the current wear factor. The standard wear amount is taken as 1 / 1000 of the cumulative wear value from new equipment to the first major overhaul, ensuring that the center value of W_cost ≈ 1.

[0072] Step 1.2: Construct the dynamic weight update equation: α,β,γ=f(electricity price, health, emergency).

[0073] Specifically, the three factors α, β, and γ are mapped to the interval [0,1], and then normalized using Softmax. The mapping process is as follows: When the real-time electricity price is less than or equal to the off-peak electricity price, f_p is mapped to 0; when the real-time electricity price is greater than or equal to the peak electricity price, f_p is mapped to 1; otherwise, f_p is linearly interpolated.

[0074] The formula for calculating the health factor is f_h = 1 – (system average RUL) / 100. The lower the average RUL, the more wear needs to be reduced, and f_h approaches 1. Construct the urgency factor calculation formula f_u=1–(remaining time until the deadline) / (maximum allowed time). The more urgent the task, the more time is needed, and the closer f_u is to 1.

[0075] The formula for calculating Softmax normalization is: exp_u=exp(k_u·f_u), exp_p=exp(k_p·f_p), exp_h=exp(k_h·f_h); α=exp_u / (exp_u+exp_p+exp_h); β=exp_p / (exp_u+exp_p+exp_h); γ=exp_h / (exp_u+exp_p+exp_h); Among them, k_p, k_h, and k_u are business sensitivity constants, which can be calibrated at the factory (e.g., 2, 2, 1.5) or corrected through online learning to ensure α+β+γ=1, with a real-time refresh cycle of 30s to 1min.

[0076] Step 1.3: Task-level cost calculation (calculate the cost of each candidate path in the scheduled task, and the one with the lower cost wins).

[0077] For each candidate path i of a feasible vehicle-associated path, calculate the task-level cost Cost_i = α·T_cost_i + β·E_cost_i + γ·W_cost_i; the scheduling engine traverses all feasible vehicle-path combinations and selects the one with the minimum Cost_i.

[0078] Step 1.4: Execution strategy for peak shaving and valley filling.

[0079] If β > 0.5 (peak electricity price), non-urgent and deferable tasks are moved to the waiting queue; for tasks that must be executed, the maximum acceleration and boost rate are reduced, and the system automatically recalculates E_cost. For example, tasks that are "non-urgent and can be postponed for ≤30 minutes" are moved to the waiting queue; for tasks that must be executed, the maximum acceleration is reduced by 20%, the boost rate is reduced by 10%, and the system automatically recalculates E_cost, which can typically reduce peak load by 8–15%.

[0080] If β < 0.2 (off-peak electricity period), idle vehicles should be returned to charging / standby stations in advance to consume off-peak electricity instead of peak electricity; within a safe range, slight overload operation is allowed (e.g., 5% overload) to bring forward some tasks to the off-peak electricity period the next morning.

[0081] Step 1.5: Digital twin feedback correction (online correction of energy consumption model and wear factor through digital twin model).

[0082] After each task, the measured energy consumption and actual wear increment are written back to the twin. The energy consumption model and wear factor are corrected online using the sliding window averaging method to ensure that the errors of E_cost and W_cost are both less than 5%.

[0083] Step 1.6: Security Boundaries and Anomaly Degradation.

[0084] Set hard upper limits for individual items: T_cost≤2.0, E_cost≤2.5, W_cost≤3.0 to prevent abnormal weights from causing device overload or timeouts. When communication is interrupted or the model fails, automatically switch back to the default strategy (e.g., the shortest time strategy) to ensure system uptime.

[0085] Step 2: Digital twin-driven device health prediction.

[0086] Digital twin models were created for each four-way car, hoist, and tunnel track, and operational data was collected in real time. This data included: motor current, voltage, and temperature of the four-way car; start-stop frequency and operating duration of the hoist; and track vibration and positioning deviation of the tunnel track. Machine learning algorithms were used to predict the remaining useful life (RUL) of key components. Figure 5 As shown, step 2 includes the following steps 2.1 to 2.7.

[0087] Step 2.1: Establish a three-dimensional digital twin framework of "component-load-operating condition".

[0088] The system is broken down into the smallest replaceable units. For example, the four-way vehicle is broken down into 4 traveling wheels, 2 servo motors, 2 gearboxes, and 4 sliding contact brushes; the hoist is broken down into 1 traction wheel, 2 chains, 1 reducer, and 1 brake; and the tunnel is broken down into a track section and a rack section every 3m, as shown in Table 1.

[0089] Table 1 Based on this, a twin class ComponentTwin is created for each smallest replaceable unit. For ComponentTwin, the attributes are {model, rated load, manufacturing date, initial cumulative wear = 0, health = 100%}; the method is update(operating data) → outputs RUL and current wear rate.

[0090] Step 2.2: Raw Data Acquisition and Edge Preprocessing. The acquired data and preprocessing methods are shown in Table 2 below: Table 2 Employing a "feature packet" mechanism: generating an 18-dimensional vector X=[timestamp, I_rms, I_cf, T_motor,E_vib1 … E_vib6, Dev_pos, N_start, Load_pct] within 1 second.

[0091] Uploaded in JSON format via MQTT (Message Queuing Telemetry Transport Protocol), single device traffic is <2 kB / s, which can be supported by 4G / NB-IoT.

[0092] Step 2.3: Establish an equivalent wear physical channel as an interpretable layer.

[0093] For each component, establish a cumulative wear equation based on "load-intensity-time": M_i(t)=∫_0^tα_i·S_i(τ)·L(τ)^β_id; Where S_i(τ) is the working condition intensity vector (vibration RMS, current peak, temperature gradient); L(τ) is the real-time load rate = measured load / rated load; α_i and β_i are material coefficients, with α=1.2 and β=1.1 for the traveling wheels and α=1.4 and β=1.3 for the motor bearings. The dimensionless cumulative wear M_i(t) is obtained by bench calibration and written into the twin ROM, and is used for long-term trend prediction.

[0094] Step S2.4: In the machine learning layer, data processing is performed through the data-driven parameter channel.

[0095] Based on historical fault records and accelerated aging experiments, a full lifecycle curve (e.g., 120 curves) is obtained. A sliding window is used to select the first 80% as the training set, and the remaining 20% ​​as the test set for the actual remaining lifespan. The machine learning model structure is as follows: Input: 18-dimensional feature sequence, length = 128 s, step size 1 s; Network: Two-layer LSTM(64,32) + Attention + Linear Regression; Output: RUL_pred (hours); Loss: Huber loss, robust to outliers.

[0096] The model is trained using training and testing sets. After training the entire model, the model with remaining lifespan is migrated to the edge. Specifically, through knowledge distillation, a 3-layer small GRU is obtained and deployed inside the box. The RUL is updated every 30 seconds, and the cloud is only responsible for hot parameter updates.

[0097] Step S2.5: Integrate prediction and uncertainty quantification, and use weighted average processing: RUL_final = w·RUL_phys + (1-w)·RUL_ml; w=0.6 (initially, high physical reliability); as the runtime increases, w will be gradually reduced to 0.3, with data-driven decision-making becoming the primary approach. Simultaneously, a 95% confidence interval is output: CI = 1.96·σ, where σ=XGBoost, representing the quantile regression results for the RUL residuals. XGBoost is an optimized distributed gradient boosting library.

[0098] Step 2.6: Calculate the health indicator HI and dynamic threshold configuration.

[0099] The health index H for each device is calculated using the formula HI = 100 × (1 – M_total / M_threshold), and the dynamic threshold is determined. M_threshold is the "equivalent hours for the first major overhaul" given by the manufacturer (e.g., 8000 h for the traveling wheel). Dynamic threshold configuration: HI>70%: Green, normal scheduling; 30%≤HI<70%: Yellow, speed limit / load limit dispatch; HI<30%: Red, generate a maintenance work order and hide the task.

[0100] Step 2.7: Lifetime-Scheduling Closed-Loop Interface.

[0101] The cloud sends the latest HI and RUL to the scheduler every 30 seconds; the scheduler adds a lifetime term Cost_life to the cost function: Cost_life = λ·(1 – HI / 100); λ is dynamically adjusted by electricity price and urgency (0.05–0.30) to achieve "more work for those with high health, less work for those with high health".

[0102] Step S2.8: Self-evolution and drift inhibition.

[0103] Feature distribution drift is detected weekly using the Population Stability Index (PSI). Retraining is automatically triggered when PSI > 0.2. After repairing or replacing parts, the code is scanned on-site for confirmation, and the M_total of the part is cleared to zero and HI is reset to 100% in the cloud, completing the digital-physical synchronization.

[0104] Steps S2.1 to S2.8 above can provide early warning of critical component failures, reduce the number of sudden downtimes, help reduce equipment-level wear variance, and extend the overall overhaul time. Edge models occupy little storage space, refresh quickly, are easy to deploy, and output standardized HI and RUL values, which can be directly called by the "multi-objective cost function" to achieve simultaneous optimization of lifetime, efficiency, and energy consumption objectives.

[0105] Step 3: Perform scheduling based on wear leveling strategy.

[0106] When allocating tasks, prioritize scheduling devices with high health or low wear to achieve wear balance across the entire system and extend the overall lifespan of the equipment.

[0107] Specifically, such as Figure 6 As shown, the steps for predicting equipment health based on the equipment digital twin model include the following: 3.1~3.7.

[0108] Step 3.1: Generate a pool of available devices.

[0109] Filtering conditions: a) RUL ≥ 72 h and HI ≥ 30%; b) No current fault alarms; c) Remaining power / bus voltage meets the requirements for the next round trip. Output: CandidateList = {c1, c2, ..., ck}. Step 2: Calculate the "Wear Leveling Index (MBI)". Step 3.2: Calculate the Wear Leveling Index (MBI): For each available device ci and target roadway si, define: MBI_j = w1·(1–HI_j / 100) + w2·(1–RUL_j / RUL_max) + w3·(Number of tasks in the last 24 hours LaneWear_si / Maximum allowed number of tasks LaneWear_max) + w4·(RencentTask_i) / (recentTask_max); where w1+w2+w3=1, with empirical values ​​of 0.4, 0.4, and 0.2 respectively. All denominators are the current maximum values ​​in the library, ensuring that MBI∈[0,1]. The smaller the MBI, the newer the device and the lighter the task, and the higher the priority it should be.

[0110] Step S3.3: Perform multi-objective comprehensive scoring based on the multi-objective composite cost function: In addition to the existing composite cost of "time-energy consumption-wear", a wear leveling term is added: TotalCost(i,si)=α·T_cost+β·E_cost+γ·W_cost+λ·MBI(i,si) λ represents the wear and tear equalization weight, which is dynamically adjusted by the real-time electricity price. Off-peak electricity period: λ=0.05 (efficiency priority); Level period: λ = 0.15; Peak power period: λ=0.25 (lifetime priority, reducing equipment overuse while shaving off peak power).

[0111] Keep α+β+γ+λ=1 and refresh every 30 seconds.

[0112] Step S3.4: Perform optimal matching and conflict resolution.

[0113] An improved bipartite graph optimal matching algorithm (Hungarian algorithm, completed instantaneously in O(n³) when n≤500) is adopted; if the Head area of ​​the same lane is occupied by two vehicles in the same time slice, the vehicle with the higher MBI is forced to decelerate for 1 time slice (10s), and the TotalCost is recalculated until there is zero conflict.

[0114] Step S3.5: Update the roadway wear online.

[0115] After each task is completed, the equivalent roadway wear increment is written back. The expression for the equivalent roadway wear increment is: ΔLaneWear = k·Load_pct·(1+Dev_pos)·Δt, where k is the track material coefficient (k=1.0 for steel rails, k=1.3 for aluminum rails), Dev_pos is the positioning deviation RMS, and Δt is the mission duration.

[0116] The cloud refreshes LaneWear_si in batches every 10 minutes and distributes it.

[0117] Step S3.6: Adaptively adjust the health threshold.

[0118] HI_th_green = 70% (Green, normal scheduling); HI_th_yellow = 30% (Yellow, speed limit / load limit); HI_th_red = 30% (Red, immediately disable the task).

[0119] Using the mean time between failures (MTBF) and mean repair cost (C_rep) statistically analyzed within the most recent overhaul cycle T_overhaul, an economic cost function is established: F(th) = λ1·(T_overhaul – T_actual)² + λ2·C_rep; Furthermore, for th∈[25%, 40%], scan with a step size of 1%, and take the minimum value of F(th) as the next cycle HI_th_red.

[0120] This forms a threshold self-evolution closed loop. Step 3.7: Closed-loop feedback and self-evolution: For example, periodically calculate the wear variance σ_wear = std(MBI_all); if σ_wear increases by more than 5% for two consecutive weeks, automatically increase λ by 0.02 until σ_wear decreases; repair and replace parts, scan the code to confirm after repair, clear the MBI component of the part to zero, and reset HI to 100% to ensure that the digital twin is consistent with the physical object.

[0121] Through steps 3.1 to 3.7 above, wear variance can be effectively reduced, local overuse can be avoided, and the overall overhaul cycle can be extended. Through dynamic adjustment of λ, more balance can be achieved when the electricity price is high and more efficiency can be achieved when the electricity price is low. The health threshold adaptive method makes the threshold change with the operating conditions without manual correction.

[0122] Step 4: Global dynamic task orchestration based on spatiotemporal decomposition.

[0123] By integrating with the WMS order forecasting module, potential tasks within the next few minutes are predicted, and idle vehicles are pre-assigned to the reserve area. The system analyzes the pickup and drop-off locations of all current tasks, intelligently merging multiple tasks within the same lane and adjusting the task order to avoid conflicts or congestion when vehicles are at the same lane end.

[0124] Furthermore, such as Figure 7 As shown, the steps 4.1 to 4.7 are included: Step 4.1: Establish a unified coordinate system for spatiotemporal grids.

[0125] Establish a spatial grid: Divide the map into cubic grids according to preset specifications. Each cargo location, aisle, hoist entrance, and charging station has a unique (x, y, z) code. For each aisle, additionally define an end area (Head) and a straight-ahead area (Body) for subsequent collision detection.

[0126] Establish a time grid: The time axis is divided according to the preset time slice size, for example, Δt=10s, forming discrete time slices of 0, 1, 2...N. T0 represents the current time, and T1~T6 represent the next 6 time slices (a total of 60s prediction window).

[0127] Step 4.2: Predict from a timeline perspective and generate a quasi-task queue.

[0128] The system retrieves the list of inventory locations expected to be delisted within the next 60 seconds from the WMS order prediction module, along with a probability P ≥ 0.6, as input. It also obtains the hourly prediction hit rate η = 85% as the historical hit rate, which is used to adjust the confidence level. A quasi-task pool is generated using a quasi-task generation algorithm, and the resulting quasi-task pool (StandbyPool, typically containing 5-15 quasi-tasks) is used for pre-deployment.

[0129] Step 4.3: Model from the perspective of spatial axes and construct a roadway conflict map.

[0130] Node: Each roadway segment + hoist entrance = graph node V_i.

[0131] Edge: If two nodes share a "shared end area" or "one-way in the same direction", then add an undirected edge E(i,j) with a weight of 1, indicating that only one car can be accommodated at a time.

[0132] Conflict matrix: Establish a 0-1 matrix C[i,j], where C=1 indicates that if i and j work simultaneously, a conflict will occur.

[0133] Step 4.4: Conduct spatiotemporal joint pre-deployment.

[0134] Using the current list of idle vehicles FreeCars={c1…ck} and the quasi-task queue StandbyPool as input, when FreeCars and StandbyPool are not empty, select the quasi-task q with the highest prediction probability; calculate the spatiotemporal path Pathi required to pre-deploy ci to q; evaluate Pathi using equipment scheduling costs (time + energy consumption + wear + MBI); perform optimal matching between ci and q to generate pre-instructions; remove matched items from both pools; output the pre-instruction queue PreCmdQueue, where each element of PreCmdQueue is {vehicle ID, target waiting grid, arrival time, reason "predicted standby"}.

[0135] Step 4.5: Real task arrives, spatial axes merge and reorder.

[0136] If there are ≤3 adjacent cargo locations in the same lane, they are combined into one "multiple pickup" task. After the combination, the tasks are executed in the order of "farthest to nearest" to reduce reversing.

[0137] The merged tasks and pre-instructions are input into the spatiotemporal conflict graph. An improved tabu search (TS) is used to find the sequence of "minimum total cost + zero conflict" within 300ms. The objective function is ΣCost_total + M·number of conflicts, where M=10. 6 Ensure hard penalties; perform neighborhood operations: swap task order, insert waiting slots, and change vehicles. Output a conflict-free task sequence FinalSeq, where each element of FinalSeq is {vehicle, cargo location sequence, start time, speed level}.

[0138] Step 4.6: Conflict micro-scheduling, avoiding head-end conflicts.

[0139] Configure the real-time location feedback cycle, for example, a cycle of 200ms. If vehicles enter the same Head area in the same 10-second time slice, micro-schedule is triggered to make vehicles with low health or non-emergency vehicles slow down (for example, by 20%) and automatically give up the Head area. If there is still a conflict, the following vehicle is ordered to temporarily stop in the Body area for one time slice (for example, 10 seconds).

[0140] Step 4.7: Perform closed-loop feedback and self-evolution.

[0141] The system calculates weekly statistics on the number of vehicle reversing at the terminal, the head idle rate, and the prediction hit rate η. If η < 80% for two consecutive weeks, the prediction window is automatically shortened to 40 seconds. If σ_wait (head wait standard deviation) increases by > 5%, λ + 0.02 is automatically increased until σ_wait decreases. The saved seconds are then used to train the prediction module, enabling self-evolution. If the decision layer detects an order prediction error probability > 30%, PreCmdQueue is immediately rolled back, releasing the vehicle return charging position.

[0142] Through steps 4.1 to 4.7 above, when the prediction hit rate is greater than 85%, it effectively reduces empty mileage and monthly electricity demand, reduces the number of terminal reversing conflicts, increases the throughput of a single lane, and the self-evolution mechanism allows the model to automatically scale with business fluctuations, eliminating the need for manual parameter tuning after deployment.

[0143] To implement the methods described in the embodiments of this application, a four-layer architecture can be used. Specifically, the four-layer architecture is shown in Table 3: Table 3 The perception layer includes hardware nodes for the four-way vehicle, the hoist, and the tunnel. Specifically, the four-way vehicle node includes current, voltage, vibration, temperature, encoder, and visual positioning; the hoist node includes traction wheel encoder, tension, vibration, and temperature; and the tunnel node includes track vibration, straightness visual sensor, and rack wear laser sensor.

[0144] The interface is standardized, and the edge compression of the 1Hz feature packet is processed using 1kHz (current / vibration). RMS, FFT and CRC checks are performed on the vehicle MCU / lane gateway. Abnormal frames (CRC errors) are immediately retransmitted. If the retransmission exceeds 3 times, a "sensor fault" flag is reported.

[0145] At the model layer, a digital twin model of the equipment is constructed, outputting HI and RUL; a dynamic energy consumption model of the warehouse is constructed, outputting the real-time E_cost baseline; a multi-objective composite cost function model is constructed, adjusting the weights of α, β, γ, and λ in real time. A closed-loop sequence number mechanism is adopted, with frame_id (i.e., the closed-loop sequence number) globally incremented. When the decision layer sends back the execution frame, it must bring back the same-source frame_id, realizing the four-loop locking of "perception-model-decision-execution".

[0146] The decision-making layer is configured with a spatiotemporal decomposition task orchestrator, wear leveling scheduler, and fine-tuning conflict resolver. The instruction frame carries a timeout (TTL) of 100 ms. If the execution frame does not receive the next frame within the TTL, it will automatically decelerate and stop. The decision-making layer log stores the most recent 10,000 frame_ids and supports fault replay.

[0147] The execution layer includes equipment nodes for the four-way vehicle, the hoist, and the track side. The four-way vehicle node includes a servo drive, a PLC, and an on-board gateway; the hoist node includes a frequency converter, an encoder, and a PLC; and the track side node includes an electronic control baffle, charging contacts, and LED status lights.

[0148] The execution layer must send back execution feedback within 100 ms; otherwise, the decision layer marks "communication lost" and triggers deceleration and stopping. If ack=2 (fault), the decision layer immediately freezes the device task, removes it from the device candidate pool, and issues a replacement vehicle.

[0149] The same frame_id is used in all four stages: perception, model, decision-making, and execution. Loss of any frame or a change in the frame_id triggers a "closed-loop anomaly" alarm. Furthermore, timestamp verification and data integrity checks are performed. The maximum allowable network jitter is ≤50 ms. If the time difference at any layer exceeds 50 ms, NTP forced time synchronization is triggered, and drift logs are recorded. A CRC check is added to each interface layer; if the check fails, the data is immediately retransmitted. If three retransmissions fail, the device is marked as "degraded," and task traffic is switched to an adjacent device.

[0150] After the four-layer architecture is processed and the interface is standardized, each layer can be upgraded independently. A single node failure will not affect the whole system. The system availability is ≥99.9%. 2. The closed-loop sequence number + CRC mechanism makes the instructions traceable and replayable.

[0151] In summary, the embodiments of this application have at least the following beneficial effects: 1. Dynamically weight and integrate the "time cost, energy cost, and wear cost", with the weight coefficients α, β, and γ adjusted in real time according to the real-time electricity price, equipment health, and order urgency, to achieve coordinated optimization of peak shaving and valley filling and wear balance.

[0152] 2. Establish a digital twin for each four-way vehicle, hoist, and tunnel track, and collect data such as current, temperature, vibration, and number of start-stop cycles in real time to predict the remaining service life (RUL).

[0153] 3. Based on the health prediction results, prioritize the allocation of tasks to equipment with high health or low roadway wear, so as to achieve balanced wear of all equipment in the warehouse and extend the overall overhaul cycle.

[0154] 4. On the time axis, combine WMS order forecasts to pre-deploy vehicles; on the spatial axis, merge tasks in the same lane and adjust their order to avoid congestion at the end of the lane.

[0155] Please see Figure 8 This application also provides a digital twin-based four-way parking garage energy efficiency and wear leveling scheduling system, which can realize the above-mentioned digital twin-based four-way parking garage energy efficiency and wear leveling scheduling method. The system includes: The first module 201 is used to construct a multi-objective composite cost function that dynamically weights and integrates time cost, energy cost, and wear cost; The second module 202 is used to predict the remaining service life and health of the equipment based on the real-time collected equipment operating parameters through the equipment digital twin model; wherein, the twin in the equipment digital twin model includes a four-way vehicle, a hoist, and a tunnel track; The third module 203 is used to generate a wear leveling index and a device scheduling cost for fused wear based on the multi-objective composite function, the remaining service life, and the device health. The fourth module 204 is used to sort the equipment in the four-way garage according to the wear leveling index of the equipment in the four-way garage and generate an equipment wear leveling scheduling strategy. The fifth module 205 is used to perform spatiotemporal decomposition of the expected tasks and global dynamic task orchestration based on the order prediction results sent by the order prediction module of the warehouse management system, combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, to obtain the target scheduling scheme, and to pre-deploy the four-way vehicle in advance.

[0156] It is understood that the content of the above method embodiments is applicable to this system embodiment. The specific functions implemented in this system embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0157] This application also provides an electronic device, which includes a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the above-described four-way garage energy efficiency and wear balance scheduling method based on digital twins. This electronic device can be any smart terminal, including a tablet computer or an in-vehicle computer.

[0158] It is understood that the content of the above method embodiments is applicable to this device embodiment. The specific functions implemented by this device embodiment are the same as those of the above method embodiments, and the beneficial effects achieved are also the same as those achieved by the above method embodiments.

[0159] Please see Figure 9 , Figure 9 The hardware structure of an electronic device according to another embodiment is illustrated. The electronic device includes: The processor 301 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this application. The memory 302 can be implemented as a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 302 can store the operating system and other application programs. When the technical solutions provided in the embodiments of this specification are implemented through software or firmware, the relevant program code is stored in the memory 302 and is called and executed by the processor 301. Input / output interface 303 is used to implement information input and output; The communication interface 304 is used to enable communication and interaction between this device and other devices. Communication can be achieved through wired means (such as USB, network cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.). Bus 305 transmits information between various components of the device (e.g., processor 301, memory 302, input / output interface 303, and communication interface 304); The processor 301, memory 302, input / output interface 303, and communication interface 304 are connected to each other within the device via bus 305.

[0160] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method.

[0161] It is understood that the content of the above method embodiments is applicable to this storage medium embodiment. The specific functions implemented in this storage medium embodiment are the same as those in the above method embodiments, and the beneficial effects achieved are also the same as those achieved in the above method embodiments.

[0162] Memory, as a non-transitory computer-readable storage medium, can be used to store non-transitory software programs and non-transitory computer-executable programs. Furthermore, memory may include high-speed random access memory, and may also include non-transitory memory, such as at least one disk storage device, flash memory device, or other non-transitory solid-state storage device. In some embodiments, memory may optionally include memory remotely located relative to the processor, and these remote memories can be connected to the processor via a network. Examples of such networks include, but are not limited to, the Internet, intranets, local area networks, mobile communication networks, and combinations thereof.

[0163] The embodiments described in this application are for the purpose of more clearly illustrating the technical solutions of the embodiments of this application, and do not constitute a limitation on the technical solutions provided by the embodiments of this application. As those skilled in the art will know, with the evolution of technology and the emergence of new application scenarios, the technical solutions provided by the embodiments of this application are also applicable to similar technical problems.

[0164] Those skilled in the art will understand that the technical solutions shown in the figures do not constitute a limitation on the embodiments of this application, and may include more or fewer steps than shown, or combine certain steps, or different steps.

[0165] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0166] Those skilled in the art will understand that all or some of the steps in the methods disclosed above, as well as the functional modules / units in the systems and devices, can be implemented as software, firmware, hardware, or suitable combinations thereof.

[0167] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “having,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0168] It should be understood that in this application, "at least one (item)" means one or more, and "more than" means two or more. "And / or" is used to describe the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0169] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of the units described above is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be an indirect coupling or communication connection through some interfaces, apparatuses, or units, and may be electrical, mechanical, or other forms.

[0170] The units described above as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0171] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0172] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes multiple instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing programs, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0173] The preferred embodiments of the present application have been described above with reference to the accompanying drawings, but this does not limit the scope of the claims of the present application. Any modifications, equivalent substitutions, and improvements made by those skilled in the art without departing from the scope and substance of the embodiments of the present application shall be within the scope of the claims of the present application.

Claims

1. A method for energy efficiency and wear balance scheduling of a four-way parking garage based on digital twins, characterized in that, Includes the following steps: Construct a multi-objective composite cost function that dynamically weights and integrates time cost, energy cost, and wear cost; Based on real-time collected equipment operating parameters, the remaining service life and health of the equipment are predicted through a digital twin model of the equipment; wherein, the twin in the digital twin model of the equipment includes a four-way vehicle, a hoist, and a tunnel track; Based on the multi-objective composite function, the remaining service life, and the equipment health, a wear leveling index and a device scheduling cost for fused wear are generated. The equipment in the four-way garage is sorted according to the wear leveling index of the equipment in the four-way garage, and an equipment wear leveling scheduling strategy is generated. Based on the order forecast results sent by the order forecast module of the warehouse management system, and combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, the expected tasks are decomposed into spatiotemporal global dynamic task orchestration to obtain the target scheduling scheme, and the four-way vehicles are pre-deployed in advance.

2. The method according to claim 1, characterized in that, The construction of a multi-objective composite cost function that dynamically weights and fuses time cost, energy cost, and wear cost includes the following steps: Determine the time cost based on the estimated task completion time and the average task completion time; Dynamically configure the time cost weighting coefficient based on the urgency of the task; The energy cost is determined based on the total energy consumption of the road section and the standard energy consumption value. The energy consumption cost weighting coefficient is dynamically adjusted according to the electricity price period; wherein, the standard energy consumption value is the measured value of the equipment completing one standard double cycle under rated load. The wear cost is determined based on the equivalent wear rate and standard wear rate of the equipment components in the four-way garage; The wear cost weighting coefficient is dynamically adjusted based on the equipment health status; wherein, the standard wear rate is calculated based on the cumulative equivalent hours from the time the new equipment is installed until its first major overhaul. A multi-objective composite cost function is constructed based on the time cost, the time cost urgency coefficient, the energy cost, the energy cost weighting coefficient, the wear cost, and the wear cost weighting coefficient.

3. The method according to claim 1, characterized in that, The process of predicting the remaining service life and health of equipment based on real-time collected equipment operating parameters using a digital twin model includes the following steps: Construct twin objects based on the four-way vehicle, hoist, and tunnel track; Based on the components of the twin object, a number of minimum replaceable units are determined; Collect equipment operation data for each of the minimum replaceable units; wherein, the equipment operation data includes load, operating intensity, and working time; The cumulative wear equation is calculated based on the equipment operating data to obtain the dimensionless cumulative wear value. A long short-term network model combining attention mechanism and linear regression is used to construct a device digital twin model based on the twin object; Using the digital twin model of the equipment, the remaining lifespan of the twin object is predicted based on the equipment's operating data. The prediction results are then weighted and averaged to obtain the remaining lifespan of the equipment. The equipment health status is calculated based on the dimensionless cumulative wear value and the expected working time.

4. The method according to claim 1, characterized in that, The step of generating the wear leveling index and the equipment scheduling cost of fused wear based on the multi-objective composite function, the remaining service life, and the equipment health includes the following steps: Based on the remaining service life and health status of the equipment, the equipment in the four-way garage is filtered to obtain usable equipment and usable lanes; For each available device, a wear leveling index is calculated based on the device health, task execution rate, and remaining service life of the available device; wherein, the target roadway is the available roadway matched with the available device; The wear and tear balance weighting coefficient is dynamically adjusted based on the actual electricity price. The equipment scheduling cost of fused wear is calculated based on the multi-objective composite function, the wear leveling index, and the wear leveling weight coefficient.

5. The method according to claim 1, characterized in that, The step of sorting the equipment in the four-way parking garage according to the wear leveling index of the equipment in the four-way parking garage and generating an equipment wear leveling scheduling strategy includes the following steps: The equipment in the four-way parking garage is screened based on its health status and remaining service life to obtain qualified equipment; The qualified equipment is sorted in ascending order of wear leveling index to obtain a candidate equipment pool sequence; An improved bipartite graph algorithm is used for optimal task matching; When the head area of ​​the same lane is occupied by multiple vehicles in the same time slice, the vehicle with the high wear leveling index is forced to decelerate, the equipment call cost is recalculated, until there is zero conflict, and the equipment wear leveling scheduling strategy is obtained.

6. The method according to claim 5, characterized in that, The method further includes the following steps: For devices whose health status and remaining service life are both below a preset threshold, a maintenance work order is triggered.

7. The method according to claim 1, characterized in that, The process involves using the order forecasting results sent by the order forecasting module of the warehouse management system, combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, to perform spatiotemporal decomposition and global dynamic task orchestration of the expected tasks, thereby obtaining the target scheduling scheme. Pre-deployment of the four-way vehicle is then performed in advance, including the following steps: A spatial grid is constructed based on the warehouse top view, and a discrete time axis is constructed based on a preset minimum time slice; wherein, the spatial grid is used to determine the position coordinates of each piece of equipment in the four-way garage; the discrete time axis is used to determine the start and end times of the expected tasks; Based on the location coordinates, nodes, edges, and conflict matrix are determined, and then a lane conflict map is constructed. For each predicted storage location, a predicted task is generated based on the task priority and the order prediction result. The predicted tasks whose prediction probability reaches the expected probability are added to the quasi-task pool to obtain the quasi-task queue. The current pool of idle vehicles is determined based on the equipment wear leveling scheduling strategy. Based on the current idle vehicle pool and the quasi-task queue, with the goal of minimizing the device call cost, the spatiotemporal path is calculated, a pre-instruction queue is generated, and the four-way vehicles are pre-deployed in advance. When the actual task arrives, according to the principle of merging tasks in the same roadway, and based on the task execution order of tasks from farthest to nearest, the actual tasks are merged to obtain an optimized scheduling scheme. The optimized scheduling scheme and the pre-instruction queue are input into the desired lane conflict map. An improved tabu search algorithm is used to reorder the optimized scheduling scheme with the goal of minimizing the scheduling cost of the second device and achieving zero conflict, thereby generating a conflict-free task sequence and obtaining the target scheduling scheme.

8. The method according to claim 7, characterized in that, The method further includes: Configure the scheduled data return period; During the timed feedback cycle, if two four-way vehicles enter the same end area at the same time slot, the vehicle with lower health status or the non-emergency vehicle will slow down; if the conflict still occurs, the following vehicle will be controlled to make a brief temporary stop in the straight-ahead area.

9. A four-way parking garage energy efficiency and wear balancing scheduling system based on digital twins, characterized in that, include: The first module is used to construct a multi-objective composite cost function that dynamically weights and fuses time cost, energy cost, and wear cost; The second module is used to predict the remaining service life and health of the equipment based on the real-time collected equipment operating parameters through a digital twin model of the equipment; wherein, the twin in the digital twin model of the equipment includes a four-way vehicle, a hoist, and a tunnel track; The third module is used to generate a wear leveling index and a device scheduling cost for merging wear based on the multi-objective composite function, the remaining service life, and the device health. The fourth module is used to sort the equipment in the four-way garage according to the wear leveling index of the equipment in the four-way garage and generate an equipment wear leveling scheduling strategy. The fifth module is used to perform spatiotemporal decomposition of the expected tasks and global dynamic task orchestration based on the order prediction results sent by the order prediction module of the warehouse management system, combined with the equipment scheduling cost and equipment wear leveling scheduling strategy, to obtain the target scheduling scheme, and to pre-deploy the four-way vehicles in advance.

10. An electronic device, characterized in that, Including the processor and memory; The memory is used to store programs; The processor executes the program to implement the method as described in any one of claims 1 to 8.

Images