A digital twin-driven real-time matching and optimization method and system for solid waste

Abstract

This invention relates to the field of digital twin technology, and more particularly to a digital twin-driven real-time matching and optimization method and system for solid waste. The method includes the following steps: acquiring source data of construction solid waste, and performing a digital twin modeling of the source data to obtain a digital twin source model; acquiring receiving data of construction solid waste, and performing a digital twin modeling of the receiving point to obtain a digital twin receiving model; matching the digital twin source model and the digital twin receiving model to obtain a solid waste matching model; and performing task splitting and fusion scheduling based on the solid waste matching model to obtain solid waste scheduling data. This invention, by constructing digital twin models of the source and receiving ends of construction solid waste, achieves refined modeling and dynamic mapping of solid waste attributes, processing capacity, and pollution load, effectively reducing the environmental diffusion risk during transportation and enhancing the feasibility of task splitting and fusion and resource carrying capacity efficiency.

Description

Technical Field

[0001] This invention relates to the field of digital twin technology, and in particular to a digital twin-driven real-time matching and optimization method and system for solid waste. Background Technology

[0002] Digital twins refer to a technological approach that utilizes multi-source data from physical entities (such as equipment, systems, and buildings) to construct a dynamic, real-time mapping model in virtual space, thereby enabling the perception, simulation, prediction, and control of the entity's state. Its core lies in achieving bidirectional interaction and synchronous updates between the physical entity and the virtual model by integrating physical models, sensor data, historical behavior, and operational logic. Currently, the resource-based transportation and treatment of construction solid waste often relies on manual experience for coarse-grained matching between the source and receiving ends, lacking systematic modeling and precise coordination of factors such as pollutant types, receiving capacity, and transportation route risks. This approach not only leads to low efficiency in solid waste transportation and high risks of pollution transfer but also easily causes congestion at receiving points and compliance overloading during peak periods. While some systems introduce static rules or geographic route planning techniques, they struggle to perform real-time updates and fusion decisions under dynamic multi-source data. Therefore, combining digital twin technology with the resource-based transportation and treatment process of construction solid waste has become a crucial issue. Summary of the Invention

[0003] To address the aforementioned technical problems, this invention proposes a digital twin-driven real-time matching and optimization method and system for solid waste, thereby resolving at least one of the aforementioned technical issues.

[0004] This application provides a digital twin-driven real-time matching and optimization method for solid waste, the method comprising:

[0005] S1. Obtain source data of construction solid waste, and perform digital twin modeling of solid waste source based on the source data to obtain a digital twin source model.

[0006] S2. Obtain data from the construction solid waste receiving end and perform a digital twin modeling of the receiving point based on the data to obtain a digital twin receiving end model.

[0007] S3. Match the digital twin source model and the digital twin receiver model to obtain the solid waste matching model;

[0008] S4. Based on the solid waste matching model, perform task splitting and fusion scheduling to obtain solid waste scheduling data.

[0009] This invention introduces digital twin modeling techniques to achieve high-precision, dynamic modeling of both the generation and receiving ends of construction solid waste. This not only improves the information symmetry between the source and receiver in the spatiotemporal dimensions but also provides a precise data foundation for matching and scheduling. Combining multi-model collaborative matching, adaptation is made based on factors such as solid waste characteristics, receiving capacity, and pollution restrictions, avoiding resource waste and path conflicts inherent in traditional static scheduling schemes. At the scheduling level, a task splitting and fusion strategy is adopted, effectively improving the transport efficiency ratio and path combinability, reducing transportation costs, and enhancing the dynamic control capabilities of multi-source heterogeneous solid waste flows. This approach is suitable for large-scale intelligent construction solid waste allocation sites.

[0010] Optionally, S1 includes:

[0011] Basic data on construction solid waste generation sources are collected to obtain basic data on solid waste sources;

[0012] Based on the basic data of solid waste sources, attribute standardization is performed to obtain the source data of construction solid waste;

[0013] Digital twin objects are constructed based on the source data of construction solid waste to obtain digital twin data of solid waste source.

[0014] Based on the digital twin data of solid waste source, a scheduling accessibility evolution simulation is performed to obtain a digital twin source model.

[0015] This invention constructs a source-end data system with a unified format and semantic tags by structurally collecting and standardizing the attributes of construction solid waste sources, thus solving the problems of heterogeneous, scattered, and spatiotemporally inconsistent solid waste data. Through the construction of digital twin objects, the system can reproduce the generation state, storage characteristics, and emission rhythm of source-end solid waste in a virtual environment in real time, providing an interactive digital mapping for simulation scheduling. Through scheduling reachability evolution simulation, the source-end model not only reflects the current state but also has a dynamic prediction function for scheduling response capabilities after task triggering, which helps improve the rationality of path planning and resource organization efficiency in solid waste matching.

[0016] Optionally, S2 includes:

[0017] Acquire data from the construction solid waste receiving end, and extract solid waste treatment features based on the construction solid waste receiving end data to obtain solid waste treatment feature data;

[0018] Pollutant limit data is obtained by extracting pollutant limits from the data received by the construction solid waste receiving end.

[0019] A digital twin object is constructed based on the data from the construction solid waste receiving end to obtain the digital twin data of the solid waste receiving end;

[0020] Based on solid waste treatment characteristic data and pollutant restriction data, the digital twin data of the solid waste receiving end is operated and evolved to obtain a digital twin receiving end model.

[0021] This invention utilizes systematic collection and in-depth data mining of construction solid waste receiving points to accurately extract their processing capacity, response characteristics, and pollutant constraints, thereby constructing a multi-dimensional and quantifiable receiving characteristic representation system. By constructing digital twin objects, static receiving capabilities are transformed into dynamically evolving digital entities, enabling the fitting and prediction of actual processing flows. Based on solid waste treatment characteristics and pollutant constraint data, operational evolution modeling is conducted, which not only enhances the receiving end model's ability to respond to task load fluctuations, pollution impacts, and resource consumption, but also provides a simulation foundation with environmental constraint awareness for pollution adaptation, carrying capacity assessment, and scheduling coordination in solid waste matching.

[0022] Optionally, S3 includes:

[0023] Based on the digital twin source model and the digital twin receiver model, cross-batch collaborative matching is performed to obtain the first solid waste matching model;

[0024] Based on the digital twin source-end model and the digital twin receiver-end model, the solid waste from polluted buildings is matched to obtain the second solid waste matching model;

[0025] Based on the first and second solid waste matching models, path risk avoidance control is performed to obtain the solid waste matching model.

[0026] This invention constructs a multi-granularity solid waste matching mechanism based on a digital twin model of the source and receiver, enabling cross-batch resource coordination and dynamic allocation, effectively improving the integrated utilization rate of construction solid waste in multi-source and multi-flow scenarios. By modeling a dedicated matching path for polluted construction solid waste and introducing pollutant migration simulation and carrying capacity constraints, the system possesses the ability to adapt to high-pollution-load scenarios, ensuring that the matching results comply with environmental regulations and processing capacity boundaries. Through path risk avoidance control, combined with dynamic adjustments based on multi-dimensional constraints such as construction interference, route congestion, and pollution diffusion in the scheduling network, the safety of solid waste transportation and the reliability of path planning are improved.

[0027] Optionally, the cross-batch collaborative matching includes:

[0028] Collaborative entity data is obtained by extracting collaborative entities based on the source model and receiver model of the digital twin.

[0029] Based on the collaborative entity data, reachability entity partitioning and splittable entity partitioning are performed to obtain reachability entity data and splittable entity data respectively;

[0030] Based on the digital twin receiver model, a collaborative batch path graph is generated for reachable entity data and splittable entity data to obtain the first solid waste matching model.

[0031] This invention introduces collaborative entity extraction, breaking the traditional static division boundaries between batches and achieving coupled modeling of the source and receiver ends in terms of spatial reachability and task structure separability. By dividing collaborative entities into reachable entities and separable entities, the system can flexibly address issues of task granularity differences and uneven batch sizes while ensuring the physical reachability of transportation paths, significantly improving the flexibility and adaptability of the solid waste scheduling process. Combined with the receiver model, collaborative batch path graph generation forms a structured graph covering multiple batches and multiple task paths, optimizing transportation path resource utilization and enhancing the overall scheduling system's reconfiguration capabilities under conditions of multi-source redundancy, critical loads, and supply-demand mismatch.

[0032] Optionally, the reachability entity partitioning includes:

[0033] Layered road network path data is obtained by performing layered road network path processing based on collaborative entity data and preset electronic map data.

[0034] Multi-level road network maps are constructed from the hierarchical road network path data to obtain multi-level road network map data;

[0035] Multi-level road network map data is processed to meet the requirements of source group assembly, resulting in path layer data;

[0036] By fusing and extracting path traffic reachability from the stacked path data, fused path traffic data is obtained.

[0037] Convolutional processing is performed on the fused route traffic data to obtain peak window data that can be spliced ​​together;

[0038] Based on the digital twin receiver model, multi-entry path assignment conflicts are performed on the peak window data that can be spliced ​​to obtain reachability entity data.

[0039] This invention introduces a hierarchical road network and multi-level graph modeling mechanism to achieve fine-grained analysis of construction solid waste transportation routes, fully considering the hierarchical differences of various traffic network structures such as main roads, branch roads, and temporary construction access roads. The demand processing of load source groups enables the system to identify task groups that can be collaboratively loaded in the early stages of scheduling, and constructs a composite path structure covering multiple task requirements by combining path layering data. By fusing path traffic accessibility extraction and convolution operations, time windows with high carrying potential in dynamic traffic flows are identified, effectively supporting the temporal organization of scheduling resources and capacity aggregation. Multi-entry path assignment conflict analysis is conducted based on a digital twin receiver model, fully considering traffic bottlenecks and task interference under multi-path access conditions at the receiver, ensuring that the selected path scheme has executability and resource coordination capabilities during peak hours, improving the load-bearing efficiency and road adaptability of solid waste transportation schemes.

[0040] Optionally, the matching of polluted construction solid waste includes:

[0041] Pollutant load characteristics are extracted based on the digital twin source-end model and the digital twin receiver-end model to obtain pollutant load characteristic data;

[0042] The receiving end pollution load data is obtained by performing pollution load characteristic data processing on the receiving end.

[0043] Based on the pollution carrying data at the receiving end, pollution migration simulation is performed on the digital twin source model and the digital twin receiver model to obtain pollution migration data;

[0044] Based on pollution migration data, the source-end model and receiver-end model of the digital twin are matched and labeled with pollution solid waste to obtain the second solid waste matching model.

[0045] The extraction of pollutant load characteristics includes:

[0046] Pollutant data is obtained by extracting pollutant data based on the digital twin source-end model;

[0047] Interaction path data is obtained by extracting interaction paths based on the source model and receiver model of the digital twin.

[0048] Pollution potential attenuation processing is performed on the interaction path data based on the pollutant data to obtain pollutant path data;

[0049] Multi-step particle migration simulation is performed on pollutant path data and preset electronic map data to obtain multi-step particle migration data.

[0050] Path overlap diffusion enhancement was performed based on multi-step particle migration data to obtain pollution exposure map data;

[0051] Based on the pollution exposure map data, a pollution migration field-receiver load simulation was performed on the digital twin receiver model to obtain pollutant load characteristic data.

[0052] This invention establishes a multi-stage modeling mechanism for the transmission and reception risks of highly polluting construction solid waste, combining pollutant load assessment with dynamic matching of carrying capacity. By extracting pollutant load characteristics, it not only quantifies the spatial-temporal diffusion trends of pollutants at the source but also incorporates the exposure risk distribution of interaction paths, providing data support for pollution path identification and control. Based on pollution carrying capacity processing at the receiving end, the system can dynamically adjust the processing strategy of the receiving unit, accurately identify the pollution load critical point, and avoid the secondary pollution risk caused by overload reception. Through pollution migration simulation, a co-evolution map of the pollution diffusion field and carrying capacity field is constructed, enabling forward-looking simulation of pollution links and identification of avoidance paths.

[0053] By constructing a dynamic migration mechanism of pollutants in complex transportation networks, quantitative analysis and path perception of the source-end pollutant propagation potential were achieved. Combining interactive path data extracted from digital twin source-end and receiver-end models, a full-cycle perspective of the pollution transmission link was constructed. Through pollution potential attenuation processing, the diffusion rate and attenuation trend of pollutants during transmission were effectively characterized, avoiding the underestimation of actual exposure risk by static factors. Through multi-step particle migration simulation, pollutant path data and map data were integrated to simulate the multi-step migration process from a particle dynamic perspective, significantly improving the perception of spatial diffusion and path tortuosity. The path overlap diffusion enhancement strategy performs exposure overlay modeling at the intersection of multiple source paths, generating more realistic pollution exposure map data. Through pollution migration field-receiver carrying capacity simulation, dynamic coupling modeling from pollution source diffusion and propagation path to receiving capacity was completed.

[0054] Optionally, the path risk avoidance control includes:

[0055] Path risk extraction was performed on the first solid waste matching model and the second solid waste matching model to obtain the first path risk data and the second path risk data, respectively.

[0056] A path risk map is constructed based on the first path risk data and the second path risk data to obtain path risk map data.

[0057] By performing avoidance processing on the path risk map data, a solid waste matching model is obtained.

[0058] This invention effectively improves the granularity of risk identification and dynamic avoidance capabilities in solid waste scheduling by introducing dual-source extraction of path risks and graph structure modeling. By extracting path risk information from the first and second solid waste matching models respectively, the system can form a full-dimensional risk perception map under the superposition of conventional transportation constraints and high-pollution paths. The path risk map construction process comprehensively considers multiple risk factors such as geographically sensitive areas, traffic bottlenecks, pollution propagation channels, and overloaded nodes at the receiving end, achieving a structured expression of the spatiotemporal evolution of transportation paths. The path avoidance processing not only supports path reconstruction based on weight adjustment, but also combines real-time scheduling status to perform dynamic path replacement and risk transfer strategy optimization, thereby ensuring that the final matching model achieves a balance between safety, ecological sustainability, and timeliness.

[0059] Optionally, S4 includes:

[0060] Task scheduling units are extracted based on the solid waste matching model to obtain task scheduling unit data.

[0061] The task scheduling unit data is subjected to task splitting criteria to obtain task splitting data;

[0062] Risk-reorganized task split data is performed to obtain task reorganized data;

[0063] Feasibility screening of task reorganization data was performed to obtain solid waste scheduling data.

[0064] This invention achieves refined decomposition and reconstruction of solid waste transportation tasks by constructing a multi-stage task scheduling mechanism oriented towards a matching model. Based on the solid waste matching model, task scheduling unit data is extracted, shifting the scheduling granularity from batch to controllable units, enhancing the system's adaptability to real-world scenarios such as complex construction sites and distributed receiving points. The introduction of task splitting criteria enables the system to dynamically segment tasks based on diverse conditions such as transportation capacity, pollution level, and time constraints, thereby improving resource utilization and reducing single-path conflict risks. The risk reorganization stage intelligently merges tasks based on path exposure, carrying capacity threshold, and historical scheduling feedback, ensuring that reorganized tasks possess controllable risk thresholds and executability. Finally, the feasibility screening process combines current road conditions, platform scheduling load, and collaborative requirements to constrain and verify the reorganized tasks, ensuring that the output scheduling scheme achieves an optimal balance between transportation efficiency, safety, and ecological constraints.

[0065] Optionally, this application also provides a digital twin-driven real-time solid waste matching and optimization system for executing the digital twin-driven real-time solid waste matching and optimization method described above, wherein the digital twin-driven real-time solid waste matching and optimization system includes:

[0066] The source-end twin modeling module is used to acquire source-end data of construction solid waste and perform source-end twin modeling of solid waste based on the source-end data to obtain a digital twin source-end model.

[0067] The receiver twin modeling module is used to acquire data from the construction solid waste receiver and perform twin modeling of the receiving point based on the data to obtain a digital twin receiver model.

[0068] The solid waste matching and modeling module is used to match the digital twin source model and the digital twin receiver model to obtain a solid waste matching model.

[0069] The task splitting and fusion scheduling module is used to perform task splitting and fusion scheduling based on the solid waste matching model to obtain solid waste scheduling data.

[0070] The purpose of this invention is as follows: S1. By structurally collecting and standardizing the attributes of source-end data, a real-time updatable source-end twin is constructed, and scheduling reachability evolution is added, making solid waste computable and predictable across multiple dimensions of "type-volume-pollution-location-time," eliminating matching biases caused by data heterogeneity. S2. At the receiving end, processing capacity and pollution limits are extracted, and a receiving-end twin model that can dynamically change with shifts, equipment status, and policy constraints is constructed, transforming the original static capacity into a time-series carrying capacity curve, exposing overload and violation risks in advance. S3. Based on the twins at both ends, cross-batch collaboration and high-pollution-specific matching are performed, and a path risk map is superimposed for avoidance control, achieving joint optimization of "compatibility-distance-capacity-pollution-route safety," significantly reducing the probability of rejection and backtracking. S4. The matching results are refined into task scheduling units, introducing splitting criteria and risk reorganization, and then outputting an executable plan after fusion feasibility screening, thereby improving the load factor and vehicle utilization rate, and reducing detour mileage and waiting time. Attached Figure Description

[0071] Other features, objects, and advantages of this application will become more apparent from the following detailed description of the non-limiting embodiments, taken with reference to the accompanying drawings:

[0072] Figure 1 A flowchart illustrating the steps of a digital twin-driven real-time matching and optimization method for solid waste is shown in one embodiment.

[0073] Figure 2 A flowchart illustrating the steps of a source-end twin modeling method according to an embodiment is shown;

[0074] Figure 3 A flowchart illustrating the steps of a receiver twin modeling method according to an embodiment is shown.

[0075] Figure 4 A flowchart illustrating the steps of a solid waste matching modeling method according to an embodiment is shown.

[0076] Figure 5 A flowchart illustrating the steps of a task splitting and fusion scheduling method according to an embodiment is shown.

[0077] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0078] The technical method of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0079] Furthermore, the accompanying drawings are merely illustrative of the invention and are not necessarily drawn to scale. Functional entities can be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0080] It should be understood that although the terms "first," "second," etc., may be used herein to describe various units, these units should not be limited by these terms. These terms are used merely to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, a first unit may be referred to as a second unit, and similarly, a second unit may be referred to as a first unit. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0081] Please see Figures 1 to 5 This application provides a digital twin-driven real-time matching and optimization method for solid waste, the method comprising:

[0082] S1. Obtain source data of construction solid waste, and perform digital twin modeling of solid waste source based on the source data to obtain a digital twin source model.

[0083] Specifically, the system collects data from multiple solid waste sources, including but not limited to physical locations such as construction sites, demolition projects, and precast component factories. The collected data fields include solid waste type (e.g., concrete, bricks, metal, insulation materials); mass per unit volume (kg / m³); particle size distribution (described in micrometers); initial storage location coordinates (using GPS positioning); generation timestamp (indicating the time of solid waste generation); estimated removal deadline; and label information indicating whether it contains pollutants (e.g., whether it contains paint residue, asbestos, or other hazardous components). The system standardizes the raw solid waste attribute data, including using Principal Component Analysis (PCA) for dimensionality reduction encoding of similar solid waste data to form attribute vectors with a unified structure. These attribute vectors include solid waste type number, mass per unit volume, particle size index, and pollutant label information. All time-related data is uniformly converted to the ISO 8601 standard time format; and all spatial coordinate data uses the WGS-84 geographic projection system. The system constructs digital twin objects based on the standardized source-end solid waste data. Each solid waste entity will be mapped to a unique source-end digital twin, bound to the following information: a unique solid waste entity number; lifecycle status (such as generation, storage, pending disposal, etc.); and real-time fields, including weather impact status (such as rainy day operation restrictions) and storage status (such as covered).

[0084] S2. Obtain data from the construction solid waste receiving end and perform a digital twin modeling of the receiving point based on the data to obtain a digital twin receiving end model.

[0085] Specifically, the system collects receiving data from construction solid waste receiving stations. The collected information includes, but is not limited to, basic information about the receiving and processing station, such as its unique identification number, maximum processing capacity (e.g., in tons per hour), open operating hours, and type of processing equipment; threshold limits for acceptable pollutants, including the maximum allowable content of volatile organic compounds (VOCs), pH range, and upper limits for the concentration of metal ions (such as lead and cadmium); the current system processing load (i.e., the amount of solid waste processed per unit time), the scheduled work schedule, and the maximum capacity information of the storage yard. The system structures and standardizes the above data, including constructing a receiving capacity vector for each receiving station, containing the processing capacity value, a list of acceptable solid waste types, the available operating time intervals, and the current load value; and constructing a pollutant limit matrix, with the maximum allowable value for each pollutant and the current remaining acceptable capacity. After data structuring is completed, the system constructs a digital twin object for each receiving and processing site. Each processing site corresponds to a digital twin receiver with a unique number, which includes the scheduling window list of the receiving site (recording the time period during which it is open and available for operation), the pollution load buffer (recording the currently received and pending pollutant loads), and the task trajectory history (recording the time, type, and source of each receiving task). At the same time, each receiving site is bound to its physical spatial location and the path graph node number it is located in.

[0086] S3. Match the digital twin source model and the digital twin receiver model to obtain the solid waste matching model;

[0087] Specifically, the system constructs a source set and a receiver set, denoted as the source-end digital twin set and the receiver-end digital twin set, respectively. Based on these two types of objects, the system constructs a bipartite graph structure composed of the source-end node set and the receiver-end node set, where the edges of the graph represent the matching relationship between the source and receiver ends. The attributes of the connected edges include material compatibility, representing the degree of material fit between the solid waste type and the receiver's processing capacity, referring to a preset material matching table; pollution matching degree, representing the degree of matching between the pollutant characteristics of the source end and the pollutant restrictions of the receiver end, calculated by the pollutant load function; operation window intersection ratio, characterizing the degree of overlap between the expected collection time of the source end and the schedulable time of the receiver end; and source-end receiver accessibility data, calculated based on the traffic accessibility between the source and receiver ends, such as normalized calculation based on the travel distance between preset electronic maps. The edge weights are calculated based on the aforementioned edge attributes.

[0088] ;

[0089] in For edge weight data, For material compatibility weighting data, For material compatibility, For pollution matching degree weight data, For contamination matching degree, The weighted data is the intersection ratio of the job windows. The intersection ratio of the operation windows. For the source end to receive reachability weight data, The system provides reachability data for the source-end receivers. Candidate paths are extracted based on edge weights, and a bipartite graph is constructed from these paths to form a collaborative matching network. For source-end solid waste objects marked as high-pollution (i.e., whose pollutant flag field is 1), the system employs a priority matching mechanism. Specifically, this includes filtering targets only from receivers with pollutant treatment qualifications and remaining pollutant receiving space; performing path pollution propagation simulations before matching; and reducing the weight of a path in the scoring function if it crosses identified sensitive areas (such as schools, residential areas, and water source protection areas), and in severe cases, eliminating the candidate path.

[0090] S4. Based on the solid waste matching model, perform task splitting and fusion scheduling to obtain solid waste scheduling data.

[0091] Specifically, the system constructs scheduling task units based on the source-receiver matching pairs determined in the solid waste matching model and in conjunction with transportation requirements. Each task unit includes key scheduling fields such as source number, receiver number, expected transport weight, and corresponding arrival time requirement. This process considers vehicle dispatching plans, mapping each matching pair to a set of specific executable transportation task units. Based on the parameters in the scheduling task units, the system sets task splitting criteria. If the transport weight of a single task exceeds the maximum load capacity of the vehicles configured by the system; or if the arrival time window of the task does not overlap with the open period of the target receiver, the original task needs to be split into multiple sub-tasks. The splitting process must meet the following three basic requirements: each sub-task must have a clearly reachable path; the target receiver of each sub-task must have receiving capacity; and the transportation cost after sub-task division should not significantly increase compared to the original task. For the multiple sub-tasks generated after splitting, the system assesses their recombination potential. When multiple tasks are detected to have the same target receiver and a high degree of path overlap (e.g., exceeding 80%) between their transport paths, the system determines that they have the potential for consolidation transportation. Before reorganization, the system performs time alignment processing, fine-tuning the start time of each sub-task to ensure they enter the loading process within the same time window, achieving task consolidation and improving transportation efficiency and vehicle utilization. For all attempted merged tasks, the system performs a fusion feasibility test to ensure the executability of the scheduling plan. This mainly includes the following three checks: assessing whether there are scheduling conflicts or resource contention in the path of the merged tasks; ensuring that the merged tasks are still within the processing time allowed by the receiving end; and for consolidation tasks involving polluting solid waste, checking whether the overall pollutant index after merging exceeds the critical threshold set by the receiving end. If all the above checks pass, the system marks the scheduled task as an executable task, includes it in the scheduling result set, and forms the scheduling data output set.

[0092] Optionally, S1 includes:

[0093] S11. Collect basic data on the sources of construction solid waste to obtain basic data on solid waste sources;

[0094] Specifically, the system targets the main sources of construction solid waste, including construction sites, demolition sites, and precast concrete plants—actual scenarios with material outflow or demolition tasks. Each target has an independently identifiable physical area and work unit, capable of independently generating solid waste. The system deploys various data acquisition devices and terminals at these solid waste source nodes, including on-site IoT sensors such as silo weighing sensors, volume monitoring sensors, and environmental composition detection sensors; mobile operation terminals such as smart mobile terminals carried by workers (supporting photo uploading and task registration); and drone scanning systems for periodically collecting images or 3D point cloud data of the storage yard. For each solid waste unit, the system collects and records the unique solid waste number, solid waste type information (classified and coded according to the national or industry five-level classification standards (such as crushed concrete, old bricks, waste steel bars, insulation boards, etc.), physical quantity parameters at the point of departure, spatial location information, timestamp information, preliminary pollution risk assessment label (preliminary identification by sensors or annotation by operators, recording whether the batch of solid waste may contain specific pollutants (such as volatile organic compounds, asbestos, oily residues, etc.)), sealing status information (recording the current solid waste stacking or transportation status, including exposed stacking, temporary covering, sealed transfer, etc.), and scheduling status code (recording the scheduling process status of the solid waste in the station, including statuses such as "awaiting loading", "allocated", "abnormal interruption", etc.).

[0095] S12. Standardize the attributes based on the basic data of solid waste sources to obtain the source data of construction solid waste;

[0096] Specifically, the system standardizes the raw data collected from the source to eliminate differences in measurement units between different collection units. Weight and volume are uniformly converted to SI units (e.g., kg, m³); coordinate projection is uniformly converted to the WGS-84 system, retaining at least 6 decimal places of precision; timestamps are converted to the UTC standard format for easy cross-regional calculations; multi-valued categorical variables (e.g., solid waste type) are mapped to one-hot codes or enumerated IDs; and pollution risk labels are Boolean mapped, i.e., Boolean values ​​are either 1 or 0.

[0097] S13. Construct a digital twin object based on the source data of construction solid waste to obtain the source digital twin data of solid waste;

[0098] Specifically, based on standardized construction solid waste source data, a corresponding digital twin object is constructed for each solid waste entity. This object dynamically reflects the state evolution, scheduling behavior characteristics, and physical parameter simulation process of the solid waste entity throughout its actual lifecycle. The twin object includes a unique digital twin ID, standardized entity data references, state transition logic (e.g., {to be collected} → {to be transferred} → {in transit} → {arrived}), interface fields (e.g., whether it supports consolidation, supported loading / unloading types (lifting / pushing / manual), whether covered transport is required), real-time fields (e.g., temperature and humidity, current weight change, current location), and risk fields (e.g., whether it is near water sources / communities requiring special treatment). A stream processing system (e.g., Kafka + Flink) is used to update the twin object's state in real time.

[0099] S14. Based on the digital twin data of solid waste source end, perform scheduling reachability evolution simulation to obtain the digital twin source end model.

[0100] Specifically, the system constructs a scheduling accessibility graph structure in the traffic space based on each solid waste source digital twin object. This includes mapping the solid waste twin object to a starting node in the traffic network; constructing a traffic graph G(V,E), where V represents traffic nodes (such as intersections and hubs), and E represents connecting edges (such as road segments, bridges, and tunnels); each edge is accompanied by multiple weighted indicators, including travel time (in minutes), distance (in meters), road congestion coefficient (range [0,1]), whether it is affected by construction closures, and whether it is during restricted hours; for path segments traversing highly polluted areas, the system sets a pollution path crossing risk value based on the building attributes of the traversing area using a preset parameter library, as one of the negative constraint indicators for the edge. The path accessibility function is calculated as follows: ,in For path reachability data, To estimate the weighted data for travel time, the unit is minutes to the power of negative one. Estimated travel time, in minutes, for selecting the shortest path from the source to the receiver. The average congestion coefficient is the weighted data. This represents the average congestion coefficient along the route, with a value ranging from [0,1]. A larger value indicates more severe congestion. This is the total risk weight data. The total risk weight for highly pollution-sensitive areas traversed along the path (such as ecological protection zones, schools, hospitals, etc.) is obtained by accumulating the risk values ​​of each segment. Path accessibility data is added to the digital twin data at the solid waste source. An intermediate path accessibility scheduling threshold is set. If the current path accessibility data is lower than the path accessibility scheduling threshold, it is considered that scheduling is not feasible. At this time, the corresponding solid waste twin object will enter the scheduling queue, and the system will re-evaluate its scheduling feasibility in the next round of accessibility evolution update.

[0101] Optionally, S2 includes:

[0102] S21. Obtain data from the construction solid waste receiving end, and extract solid waste treatment features based on the construction solid waste receiving end data to obtain solid waste treatment feature data.

[0103] Specifically, the system collects key operational data from each solid waste treatment node through multi-source information channels. The data includes SCADA systems deployed at receiving stations, transfer stations, and terminal processing centers; programmable logic controller (PLC) equipment and loading machinery control terminals connected to the production line; heat treatment systems, physical sorting equipment, and compaction devices used in the solid waste treatment line; and auxiliary data collection systems such as intelligent weighing systems, video recognition, and material recognition systems. The system extracts the following fields for each receiving point and organizes them in a structured format: receiving station number and its spatial coordinates (latitude, longitude, and elevation); maximum and current processing capacity, in kilograms per hour (kg / h), which can be dynamically updated; the type of process available at the receiving point, including but not limited to pyrolysis, physical crushing, screening, and biological treatment; the current operating status of the process line, with selectable statuses including "running," "standby," and "under maintenance"; the processing cost model, which supports pricing based on solid waste type, pollution risk level, or processing weight, and the modeling method can be linear function, piecewise function, or polynomial fitting; the daily receiving window, such as "08:00–18:00"; the current cumulative receiving volume (daily statistics); inbound logistics constraints, including height limits (in meters) and weight limits (in tons); loading and unloading method requirements, such as whether hoisting, pushing, and manual unloading are supported.

[0104] S22. Extract pollutant limits based on the data from the construction solid waste receiving end to obtain pollutant limit data;

[0105] Specifically, the system systematically identifies the acceptance tolerance and treatment boundaries for each receiving end for different types of pollutants, including the types of pollutants that are not accepted, the maximum acceptable concentration, and the classification and isolation conditions required during treatment. The system extracts pollutant restriction information by combining multi-source data and platform interfaces. By connecting to regional environmental management platforms (such as provincial hazardous waste supervision information systems), the system retrieves the legally permitted pollutant acceptance data for each site, including permitted pollutant types, maximum acceptable concentrations, and total quantity limits. Based on key process parameters of the receiving end's treatment line (such as heat treatment temperature, gas purification methods, and biodegradation treatment pathways), the system determines the acceptable range for specific pollutants at that site. For example, if the treatment equipment does not have VOC adsorption treatment capabilities, then that type of pollutant is set as unacceptable. The system analyzes the site's historical acceptance records and environmental penalty information. If the site has been penalized for accepting illegal pollutants (such as asbestos or lead-containing solid waste), the system will add a "restricted acceptance" or "blacklist" tag to that type of pollutant. During the process, the system constructs a pollutant tolerance matrix, where each row represents a type of pollutant (e.g., the first row is VOC), its corresponding maximum allowable concentration (in ppm), and whether acceptance is permitted (Boolean value).

[0106] S23. Construct a digital twin object based on the data from the construction solid waste receiving end to obtain the digital twin data of the solid waste receiving end;

[0107] Specifically, an independent digital twin object is constructed for each construction solid waste receiving terminal. Each digital twin object includes a unique identifier for the receiving terminal; a bound standardized processing feature structure, including processing technology, equipment capacity, operating time window, loading and unloading constraints, etc.; a bound pollutant restriction structure, including acceptable pollutant types, concentration limits, blacklist, etc.; a set of real-time operating status variables for the receiving terminal, including remaining processing capacity, current queuing status, historical load curve, etc.; a pollutant reception feasibility calculation function, which performs acceptability calculations on candidate tasks based on factors such as current capacity status, pollutant type, and historical violation records. The specific calculation process can be based on weighted averages or linear mapping based on the current capacity status; a scheduling linkage interface field, indicating whether it supports scheduling system integration, such as whether it supports advance reservations and whether it has load feedback capabilities; and a visualization display field for map systems or control panels, such as virtual space display and using color (green / orange / red) or saturation to reflect the remaining proportion of processing capacity or the restriction level. When the system detects that the current receiver's cumulative daily reception volume has reached 80% or more of its maximum processing capacity, it updates the status of the twin object to "restricted reception" and significantly reduces its reception capacity for new tasks. If the system detects that a candidate transportation task contains a VOC type pollution source, and the current receiver's acceptance limit has a Boolean value of 0 (i.e., receiving this type of pollutant is prohibited), the system automatically removes the task-receiver combination and does not schedule it. If the SCADA system or PLC data feedback receiver processing equipment is down, the system pushes a message to the scheduling engine via an interface, automatically triggering temporary capacity adjustments and site switching.

[0108] S24. Based on solid waste treatment characteristic data and pollutant restriction data, the digital twin data of the solid waste receiving end is run and evolved to obtain the digital twin receiving end model.

[0109] Specifically, based on the current processing capacity curve and the queued task list, the system simulates the processing capacity consumption per unit time in the future period and dynamically deducts the capacity required for the current task to predict the remaining capacity. According to existing transportation tasks and pollution types, the system calculates the expected arrival volume of various pollutants in different time periods in the future, compares it with the upper limit of pollutant tolerance at the receiving end, and forms an estimate of the remaining pollutant tolerance. This estimate is used to derive the pollutant load scoring index, reflecting the backlog trend of pollution risk. The system performs the state prediction process with a time granularity of 30 minutes, with a maximum prediction period of 24 hours, forming a set of time-series state evolution trajectories.

[0110] Optionally, S3 includes:

[0111] S31. Based on the digital twin source model and the digital twin receiver model, cross-batch collaborative matching is performed to obtain the first solid waste matching model;

[0112] Specifically, solid waste entities with geographical proximity (within a spatial range threshold), generation time proximity (generation time within an allowable error range), and material compatibility (similarity calculation based on material type (e.g., concrete, bricks, steel bars), excluding source points with significant chemical conflicts or incompatible handling methods) are extracted from multiple source models as a "collaborative entity set" to construct a batch pool. Based on the above collaborative matching relationships, the system constructs a collaborative batch graph G=(V,E), where node V represents the digital twin object of the solid waste source; edge E represents the possible collaborative paths that satisfy the matching rules, and each edge can be attached with a weight value, such as a collaborative strength score. The system uses a density-based graph clustering algorithm (such as DBSCAN or density-guided partitioning method) to partition the collaborative graph into batch units, generating a collaborative batch set.

[0113] S32. Based on the digital twin source-end model and the digital twin receiver-end model, the solid waste from polluted buildings is matched to obtain the second solid waste matching model;

[0114] Specifically, for batches marked as highly polluting at the source (such as lead-containing paint, asbestos boards, etc.), the system extracts the types, concentrations, and estimated release amounts of pollutants. For each receiver's digital twin model, the system extracts its maximum tolerable threshold for pollutants, which can be derived from equipment processing capacity, environmental permit restrictions, or historical penalty records. If the estimated release amount exceeds the maximum tolerable threshold, the pairing is automatically rejected. Combining map data and historical wind direction / drainage models, the system simulates pollutant flow paths and uses pollutant particle migration simulations (such as Gaussian particle diffusion) to estimate a heat map of pollution transfer risk. If the polluted batch is estimated to cross highly sensitive areas (such as schools, hospitals, and drinking water sources) in its path, the system prohibits pairing; otherwise, it marks it as a compliant pairing.

[0115] S33. Based on the first solid waste matching model and the second solid waste matching model, path risk avoidance control is performed to obtain the solid waste matching model.

[0116] Specifically, for each solid waste transport source-receiver combination determined through prior batch matching (first solid waste matching model) and pollution matching (second solid waste matching model), the system extracts its scheduled transport path and divides the path into several unit segments. For each unit segment, a traffic congestion index is extracted, representing the average congestion level of the segment over a historical period, which can be provided by a traffic big data platform or historical traffic flow database; an accident frequency index, representing the average frequency of traffic accidents occurring in the segment in recent years, which can be derived from a traffic management system or public security database; and a sensitive area crossing penalty, where if the segment crosses an environmentally sensitive area (such as a school, hospital, or area with a high incidence of historical pollution complaints), this value is set as a non-zero penalty factor and quantitatively marked according to the sensitivity level. Based on the above risk factors, the system constructs a weighted graph structure, where nodes are sets of path nodes, edges are sets of path segments (each segment connects two path nodes), and edge weights are the edge weights of the path segments from the node to the connected node, obtained by weighted summation based on the aforementioned indices. In the constructed weighted risk graph, the system uses... The algorithm, or Dijkstra's shortest path algorithm, calculates the minimum risk path from the source node to the receiver node and obtains its cumulative risk score. This score is the sum of the edge weights of all path segments.

[0117] Optionally, the cross-batch collaborative matching includes:

[0118] Collaborative entity data is obtained by extracting collaborative entities based on the source model and receiver model of the digital twin.

[0119] Specifically, the system sets the following collaboration conditions to limit solid waste entities that can be classified into the same collaboration batch: **Category Consistency Rule:** Collaborative entities must belong to the same solid waste category, such as concrete, bricks and tiles, wood, etc., specifically based on the material category field recorded in the digital twin source-end model; **Time Proximity Rule:** The generation time difference of collaborative entities should not exceed a preset threshold (e.g., 24 hours); **Geographic Proximity Rule:** The spatial distance between collaborative entities should not exceed a preset distance threshold (e.g., 5-20 kilometers); **Receiving Capacity Constraint Rule:** The total mass of collaborative entities must not exceed the maximum receiving and processing capacity of any target receiving point per unit time (which can be queried through the processing capacity field in the receiving twin model). The system completes the collaborative entity clustering and screening process according to the following steps: (1) Extract the basic feature tuples of solid waste from multiple source twin models, including geographical location (latitude and longitude coordinates), material type identifier, solid waste generation timestamp, and solid waste weight; (2) Execute a physical attribute-based clustering algorithm, such as K-Means or density clustering algorithm (DBSCAN), on the above tuple set to group solid waste entities that meet the characteristics of "same type, near time, near location"; (3) For each clustering result, call the digital twin receiver model to query the current available processing capacity of its corresponding receiver point. If the total mass in the cluster group exceeds the upper limit, it will be removed or split.

[0120] Based on the collaborative entity data, reachability entity partitioning and splittable entity partitioning are performed to obtain reachability entity data and splittable entity data respectively;

[0121] Specifically, the system constructs a multi-level road network model for the target area based on electronic map data. This model divides the area's roads into a main road layer, representing high-grade traffic roads with strong carrying capacity; a secondary road layer, connecting main roads to solid waste generation or treatment nodes; and a restricted / closed road section layer, including construction closures, time-limited access sections, and other areas requiring special treatment. For each pair of potential traffic paths between solid waste source and receiving points, the system calculates: ,in For entity reachability data, The shortest feasible distance from solid waste source point i to receiving point j is... For the natural constant term, The congestion coefficient is calculated based on historical traffic data and real-time road conditions, such as actual vehicle speed / speed limit. The system sets an entity accessibility threshold. The system classifies entities according to the following rules: if the entity accessibility data is greater than or equal to the entity accessibility threshold, then entity i is classified as an accessible entity, indicating that it has good transportation conditions and can be prioritized for scheduling; otherwise, the entity is marked as an entity with a path to be optimized or an inaccessible entity, and can be selected for optimized path or removed according to scheduling needs.

[0122] The system determines the physical properties and material composition of cooperating entities, triggering a split flag if any of the following conditions are met: the total mass of the entity exceeds the maximum load capacity threshold of the transport vehicle (e.g., 3.5 tons); the total volume of the entity exceeds the maximum volume threshold of the temporary storage yard at the receiving end (e.g., 2.0 cubic meters); or the entity contains two or more conflicting material components (e.g., asbestos and gypsum, which must be handled separately at the receiving end). The system constructs a table of material mutual exclusion rules based on expert experience or preset parameters. For example, asbestos must not be mixed with flammable plastics, and gypsum must not be processed in the same batch as high-cement-content concrete. If an entity contains a combination of mutually exclusive materials, a material conflict is identified. For each entity marked as splittable, the system records the reason for triggering the split, including but not limited to overload, exceeding volume limits, and conflicting material compositions.

[0123] Based on the digital twin receiver model, a collaborative batch path graph is generated for reachable entity data and splittable entity data to obtain the first solid waste matching model.

[0124] Specifically, the system combines the spatiotemporal distribution characteristics of reachable and splittable entities, using a digital twin receiver model as the target node, to construct a traffic path graph and perform batch path planning for source-end solid waste entities, forming a collaborative scheduling graph structure. With each solid waste receiver as the target node and all source-end entities to be scheduled as the starting nodes of the graph, a directed traffic graph G=(V,E) is constructed, where V represents key location points in the traffic network, such as source points, receiver points, and intersection nodes; E represents road connection edges with attributed weights. For each path edge in the graph, a cost weight function is calculated: ,in For path cost weights, For distance factor weights, The path distance weight coefficients from node i to j are obtained based on path distance normalization. For time factor weights, The weighting coefficients for estimated transportation time are obtained by normalizing the estimated transportation time. For distance factor weights, The traffic risk level weighting coefficient for the route segment is obtained based on traffic risk level normalization, accident probability and historical congestion level, or preset. For solid waste entities marked as divisible in the previous stage, the system executes the following processing logic: according to the maximum transport load, receiving yard capacity, or material conflict requirements, large-mass or multi-component solid waste entities are split into multiple sub-batches, each sub-batch meeting the scheduling criteria. The system can match the split sub-batches to different receiving terminals, or schedule them to the same receiving point in different time periods; during the matching process, the system automatically reuses the route graph structure and selects the optimal solution by comparing route costs. The system detects the overlap of receiving terminal entry nodes in all routes at a specified time; if the scheduling density in the same time window exceeds the receiving terminal capacity limit (such as vehicle queue length, unloading concurrency), it is considered a route conflict. If a conflict exists, the system replaces it with an equivalent but higher-risk alternative route; activates the scheduling delay mechanism to adjust some entities to subsequent available time windows; if the conflict cannot be resolved, the matched route is marked as infeasible. The system outputs a collaborative batch path graph model containing information such as path planning, entity allocation, and arrival times. This model, organized by collaborative batch, records the scheduling path, whether a component has split, and related arrival information for each member entity, including the collaborative batch identifier; path information for each source entity (including source point, receiver point, and a list of nodes traversed); estimated time of arrival (ETA); and if an entity is split, the split status and the independent paths of each sub-batch must also be recorded.

[0125] Optionally, the reachability entity partitioning includes:

[0126] Layered road network path data is obtained by performing layered road network path processing based on collaborative entity data and preset electronic map data.

[0127] Specifically, the system divides all roads in the electronic map data into three network layers based on their physical attributes and traffic levels: The arterial road layer consists of roads with a width of at least 10 meters, labeled as municipal main roads, urban expressways, or national / provincial highways; the secondary arterial road layer consists of roads with a width between 5 and 10 meters, allowing medium and large vehicles to pass at lower speeds; and the secondary road layer consists of roads with a width of less than 5 meters, representing internal roads within residential or industrial areas. The system extracts the attributes of each road segment based on the electronic map data. For each starting point in the collaborative entity data, the system combines the target receiving point and its road network level, and can select a heuristic search algorithm (such as...) based on real-time requirements. Algorithms) or shortest path search algorithms (such as Dijkstra's algorithm): The algorithm is suitable for multi-weighted optimal path search when traffic penalties exist (such as congestion and traffic restrictions); Dijkstra's algorithm is suitable for shortest path calculation in static graph structures. The system automatically limits the search space based on the road level of the entity. For large solid waste transportation tasks, it prioritizes limiting the search to arterial and secondary roads; for heavily loaded entities that cannot be accessed by side roads, all side road segments are excluded from the path candidate set; for small batches, combined paths connecting side roads to arterial roads can be considered. Several alternative paths are generated for each entity. The system records the node sequence (i.e., intersections of roads traversed) of each path; the total path length and estimated time; the distribution of road levels involved; and the congestion risk assessment results.

[0128] Multi-level road network maps are constructed from the hierarchical road network path data to obtain multi-level road network map data;

[0129] Specifically, the system constructs a graph structure based on each road segment in the hierarchical road network path data, formalized as a directed weighted graph. The start and end positions of each path segment are abstracted as graph nodes, and any two adjacent path segments form a graph edge. The graph edge is directed, representing the direction of road traffic. For two-way roads, two edges are established for the forward and reverse directions. Based on road level and scheduling priority, the system divides the overall graph structure into multiple layers. Each layer constitutes a multi-level road network map. The first-level layer (Level 1) contains all main road nodes and edges that meet the high-level standards; the second-level layer (Level 2) contains secondary roads connecting main roads and local target areas; and the third-level layer (Level 3) contains branch roads or temporary passages that are only accessible to light vehicles.

[0130] Multi-level road network map data is processed to meet the requirements of source group assembly, resulting in path layer data;

[0131] Specifically, the system processes the collaborative entities in each collaborative batch, extracting their assigned set of transportation path segments. Each element in the set corresponds to a sequence of path segments from the source to the receiver, including specific graph node numbers and estimated travel time information. For any two entities, the system checks whether there is an intersection of graph nodes in the path segment sets, i.e., whether there is a shared path segment; based on the travel time window data corresponding to the path segments, it determines whether there is an overlapping area of ​​possible loading time windows between the two paths. This time window is the earliest departure time and latest arrival time interval allowed by the task. For path pairs that meet the overlap conditions in both space and time, the system calculates their loading suitability using the following loading calculation function: ,in For load compatibility data, The set of segments of the path selected for task i. The set of segments of the path selected for task j. The smaller of the two path segment counts. The degree of alignment of path scheduling time. , The sign of the maximum value. for The estimated arrival time, for The estimated arrival time, Set the maximum allowable time deviation threshold for the system (e.g., 5 minutes, 15 minutes, half an hour, or one hour). This represents the number of segments in two sets that share the same path segment identifier (i.e., the same start and end node numbers). It also represents the fit data between any two sets of entities. When the value exceeds a set threshold, the system marks the paths of the group of entities as stackable paths and constructs a path stacking structure. This means that, based on shared path segments, the transportation tasks of two or more entities are merged into a unified transportation resource and scheduling track. The system outputs the identified path stacking data in the form of a structured path graph, including the sequence of shared path segments (i.e., the intersection of path segment sets); a list of merged entity IDs; and the order of path graph nodes and their estimated arrival times.

[0132] By fusing and extracting path traffic reachability from the stacked path data, fused path traffic data is obtained.

[0133] Specifically, for each shared overlapping path segment in a shared loading route, the system extracts and aggregates the following traffic accessibility-related parameters: maximum allowable load parameter, i.e., the maximum design-allowed traffic load of the current road segment, in tons (t), taking the minimum value among all shared path segments as the upper limit of the constraint; current and predicted traffic flow parameters, based on traffic sensor data, historical traffic models, or map service APIs, to obtain the current vehicle traffic flow and predict the traffic flow trend for a specific time period in the future; real-time traffic condition level parameters, using preset levels (such as low, medium, high) to identify the current or predicted traffic pressure level of the road segment and assigning numerical weights (e.g., low = 0.5, medium = 1, high = 2). Based on the above parameters, the system constructs a fused accessibility index, with the specific calculation formula as follows: ,in To integrate the traffic accessibility index of the route, This represents the average allowable load capacity for shared path segments. The estimated travel delay time is in minutes. This is a real-time traffic pressure factor derived from road condition level conversion results. The system sets an accessibility threshold; when the traffic accessibility index of the fusion path corresponding to a certain carpooling route is less than the threshold, it is considered to lack sufficient traffic accessibility. Such paths will be removed from the reachable path set and will not participate in subsequent scheduling generation.

[0134] Convolutional processing is performed on the fused route traffic data to obtain peak window data that can be spliced ​​together;

[0135] Specifically, traffic data for each route is encoded into a time-series vector, where each bit represents the congestion index at a specific moment. A sliding convolutional kernel window (e.g., 5 minutes, length 5, stride 1, kernel content [1, -1, 0, 0]) is used to extract congestion abrupt change edges and stable low-flow intervals. The convolutional kernel performs edge detection and pattern recognition on the congestion sequence within the current window, identifying congestion abrupt change segments (e.g., sudden increases or decreases) based on a preset abrupt change judgment threshold. Stable low-congestion segments, i.e., time periods where the convolution result is a local low value, are extracted as candidate time windows for load-sharing operations. Continuous low-pressure areas in the convolution result (e.g., areas that persist for 10 minutes and have a congestion coefficient less than 0.3) are marked as peak load-sharing windows. The efficient load-sharing time period information for each route is output.

[0136] Based on the digital twin receiver model, multi-entry path assignment conflicts are performed on the peak window data that can be spliced ​​to obtain reachability entity data.

[0137] Specifically, the system constructs a multi-entry structure for each receiving end. Each entry point (such as the east gate, west gate, temporary construction area entry point, etc.) is bound to corresponding constraint parameters, including but not limited to restrictions on the types of solid waste that the entry point can accept (e.g., only concrete is allowed, asbestos-containing materials are not allowed); maximum load-bearing capacity restrictions (e.g., maximum vehicle flow or mass limit per unit time); allowed entry time windows (e.g., 8:00–10:30 is a construction-only window); and current cumulative entry task volume and historical distribution information. The system performs target receiving end adaptation checks for each peak loading window. Specifically, if multiple loading entities are simultaneously assigned to the same entry point and their peak windows overlap, it is determined as time overlap; if the materials carried by the entity do not match the entry point's supported type (e.g., containing gypsum material but prohibited at the entry point), it is marked as category conflict; if there are already scheduled or en route vehicles that have caused the entry point load to exceed its set capacity threshold, it is determined as load conflict. Based on the judgment results, the system handles each potential conflict in a graded manner. Severe conflict level: such as material conflict or cumulative load seriously exceeding the limit, the path is regenerated or the current loading task is split at the entity level; Medium conflict level: such as the entry time period overlap is high but has not exceeded the threshold, a waiting mechanism is introduced or the system is switched to a backup entry point; Minor conflict level: such as the estimated waiting time is short or the system has not yet reached full load, the system can determine it as a "reachable entity" and schedule it directly.

[0138] Optionally, the matching of polluted construction solid waste includes:

[0139] Pollutant load characteristics are extracted based on the digital twin source-end model and the digital twin receiver-end model to obtain pollutant load characteristic data;

[0140] Specifically, the system extracts the following fields for each source entity: mass and volume indicators, including total mass (tons), volume (cubic meters), and moisture content (percentage); a pollutant concentration list, such as lead (Pb, ppm), cadmium (Cd, ppm), chromium (Cr, ppm), volatile organic compounds (VOC, mg / m³), asbestos markers (Boolean), pH value, and chloride ion concentration (percentage); particle size distribution data (including d10, d50, and d90 (micrometers)); and time and geographic location data (including generation time (ISO 8601 format), latitude and longitude coordinates (WGS-84 coordinate system), and transportable time window (formatted as start and end time intervals)). All pollutant concentration fields are unit-converted, unifying metals to mass fraction (ppm) and gases / vapors to mass concentration (mg / m³). For each pollutant type, the system calculates the source-end pollutant load value using the following formula: ,in The total load of entity i on pollutant k (units are automatically matched according to pollutant type). The physical unit for this source is tons for solid state and cubic meters for gaseous state. Let k be the mass concentration value of pollutant k (e.g., Pb is ppm, VOC is mg / m³). The system calculates the toxicity hazard index, expressed as follows: ,in The toxicity hazard index, For pollutant type index, Pollutant weights This represents the pollutant load value. These are reference limits.

[0141] The receiving end pollution load data is obtained by performing pollution load characteristic data processing on the receiving end.

[0142] Specifically, a carrying capacity matrix is ​​established based on pollutant load characteristic data, namely, the hourly / daily upper limit of pollutant k for site j and the current usage. Based on this matrix, recovery / decrease coefficients are constructed using historical data. ,in For the remaining available capacity, This is the maximum capacity limit. For indexing historical moments, For the current time, This represents the received pollution load. Based on the receiver configuration and pollutant attributes, the system determines whether the current source material is permitted for reception. Specific rules include, but are not limited to, prohibiting reception of material if asbestos is listed in the pollutant inventory but the site does not have an asbestos processing permit; requiring off-peak scheduling or material splitting if the expected peak value of a pollutant (such as VOC) exceeds the hourly limit at the receiving point; and requiring pretreatment before acceptance if the pH value exceeds the receiver's permissible range.

[0143] Based on the pollution carrying data at the receiving end, pollution migration simulation is performed on the digital twin source model and the digital twin receiver model to obtain pollution migration data;

[0144] Specifically, for each source-receiver pair, the system generates a set of multiple candidate transportation paths in the digital twin scenario. The transportation area is divided into equally spaced two-dimensional grid cells (e.g., 50 meters). A grid set (50 meters) is constructed, and sensitive areas, including schools, hospitals, and water source protection areas, are marked within it. The system uses a Gaussian plume or discrete convection-diffusion model to simulate the diffusion process of pollutants in the air. The migration equation is: ,in The function of pollutant concentration The function of pollutant concentration For wind speed vectors, For concentration gradient, Where is the diffusion coefficient. For the source term release rate, Let be a position function, and let be the location of the pollution source at a certain time t. , It is a spatial position vector, that is, the observed spatial position (two-dimensional or three-dimensional). The system uses an explicit time-based simulation mechanism, with each simulation step set to a specific time interval (e.g., 1-5 minutes), to dynamically update the vehicle's trajectory, pollutant release points, and diffusion field status, thereby obtaining pollution migration data.

[0145] Based on pollution migration data, the source-end model and receiver-end model of the digital twin are matched and labeled with pollution solid waste to obtain the second solid waste matching model.

[0146] Specifically, after simulating the migration path and risk distribution of pollutants, the system combines the pollutant carrying capacity of the receiving end, the pollution output intensity of the source end, and the migration risk level on the path. Based on the preset environmental compliance rules and sensitive area avoidance strategies, the system conducts a multi-factor operability assessment and labeling of the path combination between each pair of source solid waste entities and receiving end resource units. The labeling results include matching feasibility level, pollutant category suitability, time window matching degree, and path risk level.

[0147] The extraction of pollutant load characteristics includes:

[0148] Pollutant data is obtained by extracting pollutant data based on the digital twin source-end model;

[0149] Specifically, pollutant data is extracted based on the digital twin source-end model, including solid waste type labels, composition parameters, initial mass and volume parameters, particle size distribution parameters, and the time of solid waste entity generation or reporting.

[0150] Interaction path data is obtained by extracting interaction paths based on the source model and receiver model of the digital twin.

[0151] Specifically, based on an electronic map engine (such as accessing the Gaode Map API, Baidu Map service, or deploying a locally developed GIS engine), for each pair of source and receiving entities, the system extracts a set of paths from the map service that meet the following conditions as candidate interaction paths from the source to the receiving end, according to the loading requirements, road traffic level, and traffic supervision restrictions of the current solid waste dispatching task: capable of accommodating transport loads (load limits meet vehicle tonnage requirements); avoiding closed / restricted areas; the path travel time is within the transport time window set by the dispatching task (e.g., 08:00–20:00); and the total path length and duration are within the allowable range.

[0152] Pollution potential attenuation processing is performed on the interaction path data based on the pollutant data to obtain pollutant path data;

[0153] Specifically, considering the migration characteristics of pollutants in different forms, such as gaseous pollutants (e.g., volatile organic compounds, VOCs), an exponential decay model is used to simulate the natural decay trend of concentration over time. ,in Let be the pollutant concentration at time t. The initial concentration (the concentration at the source term or in the entry path segment). For the natural index term, The attenuation coefficient is set empirically based on volatility / photolysis reaction. The propagation time of the current segment is given. For particulate pollutants (such as construction dust and metal particles), the half-life is derived using settling velocity and volatile parameters to construct a segmented decay estimation model. Based on particle diameter, density, and settling velocity and resuspension probability under the given meteorological environment, a settling-residue function is formed. The system divides each interaction path into several continuous equidistant segments and calculates the potential decay value of pollutants in each segment. The decay value of each segment is estimated using the following formula: ,in This represents the potential decay value of pollutants in the k-th path segment. The attenuation coefficient related to the pollutant material. Let k be the length of the k-th path segment. For average transport speed, This represents the local environmental sensitivity weighting coefficient. These are risk factors for residual pollution in localized areas (dimensionless or linked to pollution level tables).

[0154] This invention provides a formula for calculating the natural decay trend of concentration over time. This formula is based on the fact that gaseous pollutants undergo natural processes such as diffusion, photolysis, and oxidation during propagation, and their concentration exhibits an exponential decay trend over time. The formula originates from the first-order kinetics of pollutant concentration, and its basic form is: ,in Let be the pollutant concentration at time t. The propagation time of the segment in which it occurs. Let be the attenuation coefficient of the pollutant per unit time. This coefficient is set based on empirical parameters such as the pollutant's volatility, photolysis rate, and oxidation rate, and represents the rate of concentration change of the pollutant per unit time, which is proportional to the current concentration. Solving this ordinary differential equation yields the closed-form solution described above. The formula accurately reflects the physical behavior of volatile gaseous pollutants during propagation, such as natural diffusion and photolysis, enhancing the scientific rationality of the model. A clear concentration change trend can be constructed using only three variables: initial concentration, propagation time, and attenuation coefficient, facilitating parameter calibration and on-site deployment. This exponential model can be embedded in path maps or pollution migration networks to achieve segmented attenuation superposition and construct a pollution distribution map. This invention provides a formula for calculating the attenuation value of each path segment. This formula is applicable to both gaseous and particulate pollutants, reflecting the intrinsic properties of materials through volatilization / photolysis parameters or sedimentation / residue parameters, respectively, thus improving the model's universality. This formula consists of two parts: one is the path propagation attenuation term. This section expresses the "path dwell time". "Intrinsic decay rate of pollutants" constitutes the first-order expansion result of exponential decay, in which... and The "path retention effect of velocity control" is expressed as a function of time, because at the microscopic level, pollutant concentrations decay exponentially over time. For short time intervals (or under weak decay conditions), take its first-order Taylor expansion: Divide the path into paragraphs (This can also be represented as the kth segment mentioned above), the propagation time for each segment is... The first is the dominant term, obtained by substituting the numerical values ​​into the above formula; the second is the environmental impact adjustment term. This section represents a modification to the main term, reflecting the amplification or inhibition effect of pollution attenuation under environmental conditions. The pollution residual risk factor for a local area is represented by a normalized value [0,1], indicating the residual accumulation effect caused by factors such as low wind speed, high obstruction, and dense population. By estimating the attenuation value for each sub-path segment, a complete pollution path attenuation curve can be generated by superimposing these segments. The formula is in linear combination form, requiring no iterative solution, making it suitable for rapid deployment and dynamic updates in large-scale path networks.

[0155] Multi-step particle migration simulation is performed on pollutant path data and preset electronic map data to obtain multi-step particle migration data.

[0156] Specifically, the electronic map data of the covered area is rasterized. Each raster cell is defined as a square grid cell with a side length of 50 meters. Particle release points are set at the midpoint or representative location of each path segment; the corresponding pollutant release mass is the residual pollutant mass obtained in the previous attenuation processing step. The pollutant release mass of each segment is proportionally converted into a number of simulated particles, set according to the minimum particle mass unit; at any grid location and at any time, the pollution concentration is expressed as: ,in Let be the pollutant concentration at time t and location (x, y). For the initial release mass, Pi The diffusion coefficient is defined based on empirical data from examples, and the unit is [unit missing]. , The diffusion time (from the start of release to the current time). For diffusion thickness, It is an exponential function. The x-coordinate of the current position. The x-axis represents the location of the pollution source release. The vertical coordinate of the current position. The vertical coordinate represents the location of the pollution source release. The time step is set to (e.g., 1 minute to 5 minutes). The system updates the pollution concentration value of each grid cell at each moment and records the concentration accumulation or threshold triggering behavior. Each update process is based on the particle diffusion results of the previous moment, forming continuous temporal migration process data.

[0157] This invention provides a method for calculating pollution concentration. To simulate the multi-step spatial migration process of pollutants in a path network, a two-dimensional Gaussian diffusion simulation is performed to express the concentration distribution of pollutants spreading from the release point to surrounding grid cells in an open environment. This diffusion model can simulate the impact range of each pollution source spreading outwards, providing support for pollution impact area assessment, risk warning, and site monitoring. The formula consists of two parts: one is the intensity term. The intensity term is controlled by the pollution intensity level. The overall amplitude is obtained by using the diffusion coefficient, time coefficient, and diffusion thickness (i.e., the thickness of the vertically influential layer; for surface propagation, this can be set to the near-surface thickness (e.g., 1.5-2m)). The second is the exponential term. This determines the shape and boundaries of the spatial distribution. The model in... , A concentric circle concentration map is constructed with the diffusion distance as the radius, suitable for local diffusion under the assumption of a uniform surface. An exponential function ensures that the concentration decays rapidly to near zero, possessing physical plausibility and boundary closure. The unit combination of all parameters guarantees the result is in concentration units (kg / m³), meeting the engineering requirements of pollution simulation. Particle concentrations at multiple release points can be directly summed without solving partial differential equations.

[0158] Path overlap diffusion enhancement was performed based on multi-step particle migration data to obtain pollution exposure map data;

[0159] Specifically, the system has obtained pollution paths. After multiple steps of particle migration simulation, each path generates corresponding pollution particle concentration migration data. The contribution weight of each path is set based on its decayed residual pollution load, which can be calculated based on the total mass of pollutants in the path segment decay model. For example, the weight can be set as follows: ,in For the first Pollution contribution weight of each path To determine the residual quality of pollutants at the endpoint, This represents the total number of all paths involved in the overlay. For path index variables, Let be the residual mass of pollutants at the endpoint of the j-th path. The system weights and overlays all particle migration data on a pre-set electronic map and structures the graph to obtain the total pollution exposure distribution map, i.e., the total pollution concentration on any grid cell is defined as: ,in The total pollution exposure concentration distribution map Pollution concentration values ​​at grid points For path index variables, This represents the total number of paths. For the first Pollution contribution weight of each path For the first The path is The concentration of pollutant particles at the location.

[0160] This invention provides a formula for calculating the contribution weight of each pollutant path within the same road segment. The weight in this formula is derived from the normalization of the residual pollution load at the path's endpoint, characterizing the contribution intensity of each path to the overall pollution exposure at a single location. This invention also provides a formula for calculating the total pollution concentration on any grid cell. This formula systematically superimposes the pollution contributions of multiple paths to form an overall pollution exposure map, avoiding the local bias caused by single-path analysis. In practice, each path has an independent pollution diffusion area on the electronic map; release points from different paths do not interfere with each other, and the diffusion of their pollutant particles follows a linear superposition principle; the pollution contribution intensity of each path is positively correlated with its endpoint residual mass. Derived from particle migration extrapolation processes (such as Gaussian diffusion simulations), the first... The invention constructs the spatial range and concentration distribution of pollution along different paths, specifically to ensure the consistency of time scales in the diffusion data of different paths. At the same time, the cumulative concentration or weighted average concentration within a unified reference time or a specified time window is used to characterize the stable pollution exposure value of path m to spatial point (x,y), thereby avoiding the time series deviation caused by the difference in release timing of different paths; Characterize the contribution intensity of each pathway to overall pollution exposure; obtain This represents the cumulative effect of multiple pollution sources at a given point, and is a weighted average; it reflects the enhanced diffusion of overlapping pollution sources, meaning that concentrations significantly increase when multiple pathways converge in the same area. The values ​​of each pathway... It can be pre-calculated and cached, eliminating the need for re-simulation when adjusting weights, thus improving system computational efficiency; its mathematical structure meets the requirements of distributed processing and is suitable for large-scale map rendering or edge computing deployment.

[0161] Based on the pollution exposure map data, a pollution migration field-receiver load simulation was performed on the digital twin receiver model to obtain pollutant load characteristic data.

[0162] Specifically, the system maps the two-dimensional grid cells defined in the pollution exposure map to specific physical sub-units in the digital twin receiver model. This mapping is established based on the internal structure of the receiver, ensuring a one-to-one or many-to-one correspondence between each grid area and key functional areas (such as treatment workshops, chemical pools, ventilation points, etc.) in the receiver model. For each mapped key unit, the system sets an upper limit for the acceptable pollutant exposure concentration within a unit time period. The system traverses each grid in the exposure map; if the concentration value of a grid exceeds the upper limit of pollutant exposure concentration, the exposure period is marked as "unreachable," indicating that the unit cannot normally accept new pollutant tasks during this period. For receiving units that do not exceed the exposure threshold, the receiving unit... The current theoretical carrying capacity is If the pollution exposure value of its associated grid cell is The system then dynamically calculates its adjusted bearing capacity using the following linear deduction model: ,in For receiving unit pollutants At any moment Adjusted load-bearing capacity For receiving unit pollutants At any moment Theoretical carrying capacity The attenuation coefficient of the effect of exposure concentration on carrying capacity. This represents the total concentration of pollutants at grid position (x,y) in the exposure map.

[0163] This invention provides a linear exposure attenuation calculation formula for dynamically adjusting pollutant load carrying capacity based on the linkage between a pollution exposure map and a digital twin receiver model. This formula maps the pollution exposure map to the functional areas of the receiver, establishing a coupling path between pollution migration simulation and structural response simulation. The system automatically reduces the task carrying capacity of each sub-unit based on real-time pollution exposure concentration, enabling dynamic task scheduling based on exposure status. If the exposure concentration exceeds a preset threshold, the area is automatically set to a "no new pollutants allowed" state, enhancing the system's safety constraints under high-pollution conditions. It is determined by factors such as the receiver structure, equipment type, and pollutant treatment capacity; it is the upper limit of the system's load under ideal conditions. This indicates the degree to which a unit concentration reduces the receiving capacity; the formula uses a linear subtraction structure, which is beneficial for engineering implementation.

[0164] Optionally, the path risk avoidance control includes:

[0165] Path risk extraction was performed on the first solid waste matching model and the second solid waste matching model to obtain the first path risk data and the second path risk data, respectively.

[0166] Specifically, the system extracts the first solid waste matching model respectively. Matching model with the second solid waste The entire set of paths in the dataset. For each path... The system divides it into several sub-segments. The risk of each sub-segment is then quantified. The risk value of each path segment is calculated using the following weighted model: ,in For path segment The risk value, For risk factor index, The total number of risk factors supported by the system. The corresponding risk factor weights are set during training based on factors such as the region's historical scheduling failure probability, legal and regulatory sensitivity levels, and operational security standards. For the k-th type of risk factor in the path segment The function to retrieve values ​​on, for example Traffic delay coefficient (based on real-time congestion levels); Sensitivity to air pollution (e.g.) Concentration level); Mark high-risk areas (such as water crossings, construction sites, and water sources).

[0167] This invention provides a weighted calculation method for the risk value of each path segment, wherein the system divides the path segment into several sub-segments at fixed distance intervals (or according to node distribution characteristics). And for each sub-segment, historical experience data (such as the risk of traffic delays during congested periods, and sensitivity to air pollution) are used. The risk is quantified by weighting the concentration level and whether the route crosses an environmental control zone (i.e., high-risk areas marked by water, construction, or water sources) using preset weights. This formula achieves a refined characterization of path-level risk, facilitating the introduction of risk perception capabilities into solid waste dispatching route planning; it is also easy to expand and train, as the weight parameters can be fitted from historical data, set by expert experience, or dynamically learned, providing strong flexibility.

[0168] A path risk map is constructed based on the first path risk data and the second path risk data to obtain path risk map data.

[0169] Specifically, the system first extracts turning points, key scheduling locations, and grid boundary intersections from all path segments in the first and second path risk data to construct a node set for the graph. Each node includes its associated path, coordinates, and whether it is an entrance to a sensitive area (e.g., a school or residential area). Based on this node set, and according to path structure or geographic layer constraints, the system identifies node pairs with direct communication relationships, forming an edge set for the graph. This edge set includes distance, path segment risk, current congestion delay, and pollution exposure. A multi-dimensional attribute graph representation or a multi-layer graph is then constructed.

[0170] By performing avoidance processing on the path risk map data, a solid waste matching model is obtained.

[0171] Specifically, based on the constructed path risk graph, new avoidance optimization paths are generated; a maximum allowable threshold R_max for path risk is set; if a path R(path) > R_max, it is considered a high-risk path and enters the re-avoidance optimization process. Multi-objective approach is used. Ant colony optimization (ACO) or deep reinforcement learning (DRL) scheduling strategy networks are used, with the optimization objective set as minimizing total path risk; ensuring path length or time constraints; and prioritizing avoidance of sensitive areas / pollution spread zones.

[0172] Optionally, S4 includes:

[0173] S41. Extract task scheduling units based on the solid waste matching model to obtain task scheduling unit data;

[0174] Specifically, the input consists of valid matching triples that have passed the screening in the solid waste matching model: ,in Indicates the i-th solid waste source; This represents the j-th receiving and processing terminal; This represents the transportation path between the source and receiving points; it also includes the matched material type, available transportation capacity, pollution load information, and scheduling time window. The system groups the matching triples according to the following key-value combinations as the smallest execution granularity for task scheduling: ,in Number the solid waste source stations. Number the receiving station. It is classified as solid waste (such as construction waste, heavy metal waste, volatile organic compounds, etc.). As a unique identifier for the transportation route, For each allowed arrival time window, a separate TEU entity is generated for each packet. The system calculates the estimated arrival time based on the digital twin simulation results of the path and the receiver, and obtains the task scheduling unit data based on the aforementioned integration.

[0175] S42. Perform task splitting criteria on the task scheduling unit data to obtain task splitting data;

[0176] Specifically, when a single Task Scheduling Unit (TEU) cannot be completed within a single transport mission due to factors such as capacity, pollution limitations, or time windows, it is split into a set of sub-task units that meet the execution conditions through multi-dimensional criteria. If the total mass or volume of solid waste exceeds the vehicle's carrying capacity, it must be split. The calculation method is as follows: ,in The number of task breakdowns under the vehicle load criterion. It is a function with maximum value. The mass and volume of solid waste corresponding to the task scheduling unit. The vehicle's rated load capacity, The total volume of solid waste. The vehicle's rated loading volume, This is for rounding up. When the pollution handling capacity of the receiving end is constrained, the pollution load should be divided according to the remaining carrying capacity: ,in This represents the number of partitions under pollution load constraints. Operate for maximum value (contaminant category). Let k be the load value of the pollutant to be treated. Let be the remaining processing capacity of receiver j for pollutant k at time t. If, within a given arrival time window, the maximum receiving capacity of the receiving point (such as the gate or unloading point capacity) is insufficient to complete a single scheduling task, then it needs to be split: ,in The number of splits under the time window constraint. The total mass of solid waste corresponding to the task unit. The gate's processing capacity per unit time (e.g., tons / hour). This is the allowed time window width. If the matched path contains a bridge or passage with weight limits, and the TEU quality exceeds the weight limit of any edge of the path, then splitting should be triggered. Record the following at this time: ,in The number of splits under path weight limits. For solid waste quality, This is the minimum weight limit for all paragraphs in the path. Let be the weight limit value for any edge segment.

[0177] This invention addresses the real-world problem that task scheduling units (such as a single solid waste transportation task) may not be able to be fully executed under constraints such as resources, routes, or time windows. It constructs a multi-dimensional splitting criterion model, calculating the minimum feasible number of subtasks to be split based on capacity limitations, pollution load limitations, arrival time window limitations, and route weight limits. For example, this invention provides a carrying capacity-limited splitting formula. If either mass or volume is overloaded according to this formula, the more severe overload will be prioritized for splitting, ensuring the vehicle can fit. This invention provides a time-window receiving limit splitting formula. If the task is too large, the time window too short, or the gatekeeper efficiency too low, it must be split into multiple sub-tasks and scheduled at different times. This invention provides a path weight limit splitting formula. If the weight limit of a bridge or passage in the solid waste transportation route is lower than the task load, it must be split in advance to prevent impassability. A single transportation or processing task must meet the following conditions: transportation is feasible, reception is not overloaded, arrival is not congested, and the route does not exceed limits. If any condition is not met, the task must be split (iteratively splitting according to fixed parameters until the condition is met) to reduce the unit load. Each dimension is considered a "hard constraint," and the maximum allowable load is calculated using the corresponding resource capacity. If the actual load exceeds the capacity, the minimum number of splits required to complete the task must be calculated → the number of splits. Rounding up ensures that the task can actually be completed. The number of splits calculated based on all criteria is summarized, and the maximum value is taken to ensure that all constraints are met; a corresponding number of subtasks can be generated and handed over to subsequent scheduling.

[0178] S43. Perform risk-based reorganization on the task splitting data to obtain task reorganization data;

[0179] Specifically, the system determines reorganization by constructing a task fusion feasibility graph, where graph nodes represent each sub-TEU to be merged; graph edges represent potential merging relationships between two sub-TEUs that meet fusion conditions. A mergeable edge is created for two sub-TEUs when both sub-TEUs simultaneously meet all of the following conditions: there is sufficient path sharing between the path segments used by the two sub-TEUs, i.e., their path overlap must satisfy: ,in The preset path overlap threshold is used (e.g., 0.6). The difference in estimated arrival times (ETAs) between the two sub-TEUs should be less than the synchronization tolerance (the maximum allowable range of time differences between different tasks arriving at the receiver). The receiver inlets of the two sub-TEUs should be the same or have switchability (i.e., they are alternative inlets to each other and will not cause path conflicts or congestion) to meet the spatial conditions for merging and unloading. The total load of various pollutants after merging the two sub-TEUs must not exceed the remaining pollution carrying capacity of the receiver at the corresponding time. ,in For the merged sub-tasks on pollutants Total load, For the receiving end pollutants exist The remaining processing capacity at any given time. The total combined mass must not exceed the rated load capacity of the available vehicles. ,in For the total mass after the merger, Define the vehicle's rated load capacity. Evaluate all mergeable edges and select merging schemes with positive returns: ,in For positive return data, To reduce scheduling mileage through merging, The weighted data for the increase in risk is expressed in km / risk coefficient, ranging from 0.3 to 0.6, or obtained through regression calculation based on empirical values. This represents the increase in path risk after the merger. The weighted data for the waiting time increment is expressed in km / waiting time and ranges from 0.1 to 0.5, or can be obtained through regression calculation based on empirical values. To account for the increased waiting time caused by queue changes after merging, the system performs checks on all queues. Candidate edges are prioritized and merged step by step using the maximum matching algorithm or a heuristic greedy strategy.

[0180] This invention aims to optimize the execution efficiency of task splitting in solid waste transportation scheduling. It proposes a reorganization calculation based on a task fusion feasibility graph and a benefit-driven reorganization criterion, improving overall transportation efficiency while ensuring scheduling feasibility. By merging similar sub-tasks, the utilization rate of vehicle load and receiving end capacity is increased, reducing resource waste. Parallel scheduling of task pairs with overlapping paths and spatiotemporal proximity achieves mileage savings and task merging. Only when all constraints are satisfied simultaneously does the system establish a merging edge for that task pair in the task fusion graph. This invention provides a positive benefit evaluation formula for merging edges, which states that benefit = cost savings - risk cost - waiting cost. If the calculation result is positive, the merging is worthwhile. A graph structure is used to model the merging possibility, ensuring that all merging actions comply with scheduling, path, and spatial constraints. A multi-factor benefit evaluation function is constructed using the net benefit concept from economics to achieve a joint balance of cost, risk, and time. The merging benefit comes from the total scheduling distance of the original two task routes being greater than the new task route after merging; risk and waiting time, as significant negative impacts, are controlled through weighted penalty terms. The task fusion relationship graph is constructed based on a graph theory model, which facilitates the introduction of optimization algorithms such as maximum matching and greedy strategies to quickly find the optimal merging scheme.

[0181] S44. Perform feasibility screening on the task reorganization data to obtain solid waste scheduling data.

[0182] Specifically, the system performs the following four types of hard checks for each task to be scheduled. All checks must pass before proceeding to the scheduling plan generation stage. These checks include: the assigned vehicle's load, body volume, axle load, and dimensions must be within vehicle manufacturing specifications and road control limits; the estimated arrival time (ETA) of each task must fall within the open time window set by the receiving gate; the system assesses the gate's throughput capacity based on the task concentration within the ETA time window: the total mass of all tasks within the ETA time window must not exceed the gate's processing limit per unit time; the material categories and pollutants included in each type of transportation task must match the whitelist configured by the receiving end or gate to avoid entry restrictions due to unauthorized substances; the pollutant load of all transportation tasks must not exceed the receiving end's remaining pollution carrying capacity at the ETA time to ensure that processing capacity is not exceeded; if the transportation task contains specially regulated substances (such as asbestos and volatile organic compounds, VOCs), it is necessary to verify whether its environmental permit status has been granted; the risk level and violation probability of the scheduling path must not exceed the preset limit to ensure that the task is safe and controllable. The deadline for each task must not be later than the deadline agreed upon by the customer. If the system has integrated driver scheduling data, it must also ensure that the drivers' working hours and rest intervals comply with traffic safety regulations and that scheduling does not exceed the allotted time. Tasks that pass the above hard checks will be automatically generated by the system with scheduling instructions containing the following elements: vehicle assignment, loading and unloading sequence arrangement, and scheduling timetable construction.

[0183] Optionally, this application also provides a digital twin-driven real-time solid waste matching and optimization system for executing the digital twin-driven real-time solid waste matching and optimization method described above, wherein the digital twin-driven real-time solid waste matching and optimization system includes:

[0184] The source-end twin modeling module is used to acquire source-end data of construction solid waste and perform source-end twin modeling of solid waste based on the source-end data to obtain a digital twin source-end model.

[0185] The receiver twin modeling module is used to acquire data from the construction solid waste receiver and perform twin modeling of the receiving point based on the data to obtain a digital twin receiver model.

[0186] The solid waste matching and modeling module is used to match the digital twin source model and the digital twin receiver model to obtain a solid waste matching model.

[0187] The task splitting and fusion scheduling module is used to perform task splitting and fusion scheduling based on the solid waste matching model to obtain solid waste scheduling data.

[0188] Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended application documents rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of the equivalents of the application documents be incorporated into the invention.

[0189] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features of the invention herein.

Claims

1. A digital twin-driven real-time matching and optimization method for solid waste, characterized in that, The method includes: S1. Obtain source data of construction solid waste, and perform digital twin modeling of solid waste source based on the source data to obtain a digital twin source model. S2. Obtain data from the construction solid waste receiving end and perform a digital twin modeling of the receiving point based on the data to obtain a digital twin receiving end model. S3. Perform cross-batch collaborative matching based on the digital twin source-end model and the digital twin receiver-end model to obtain the first solid waste matching model; perform matching of polluted construction solid waste based on the digital twin source-end model and the digital twin receiver-end model to obtain the second solid waste matching model; perform path risk avoidance control based on the first solid waste matching model and the second solid waste matching model to obtain the solid waste matching model. S4. Based on the solid waste matching model, perform task splitting and fusion scheduling to obtain solid waste scheduling data; The cross-batch collaborative matching includes: Collaborative entity extraction is performed based on the digital twin source-end model and the digital twin receiver-end model to obtain collaborative entity data; reachability entity division and splittable entity division are performed based on the collaborative entity data to obtain reachability entity data and splittable entity data respectively; collaborative batch path diagram is generated based on the reachability entity data and splittable entity data according to the digital twin receiver-end model to obtain the first solid waste matching model; The reachability entity partitioning includes: Based on the collaborative entity data and the preset electronic map data, layered road network path processing is performed to obtain layered road network path data; multi-level road network map construction is performed on the layered road network path data to obtain multi-level road network map data; load source group demand processing is performed on the multi-level road network map data to obtain path stacking data; fusion path traffic accessibility extraction is performed on the path stacking data to obtain fused path traffic data; convolution processing is performed on the fused path traffic data to obtain loadable peak window data; and multi-entry path assignment conflict is performed on the loadable peak window data based on the digital twin receiver model to obtain accessibility entity data.

2. The method according to claim 1, characterized in that, S1 includes: Basic data on construction solid waste generation sources are collected to obtain basic data on solid waste sources; Based on the basic data of solid waste sources, attribute standardization is performed to obtain the source data of construction solid waste; Digital twin objects are constructed based on the source data of construction solid waste to obtain digital twin data of solid waste source. Based on the digital twin data of solid waste source, a scheduling accessibility evolution simulation is performed to obtain a digital twin source model.

3. The method according to claim 1, characterized in that, S2 include: Acquire data from the construction solid waste receiving end, and extract solid waste treatment features based on the construction solid waste receiving end data to obtain solid waste treatment feature data; Pollutant limit data is obtained by extracting pollutant limits from the data received by the construction solid waste receiving end. A digital twin object is constructed based on the data from the construction solid waste receiving end to obtain the digital twin data of the solid waste receiving end; Based on solid waste treatment characteristic data and pollutant restriction data, the digital twin data of the solid waste receiving end is operated and evolved to obtain a digital twin receiving end model.

4. The method according to claim 1, characterized in that, The matching of polluted building solid waste includes: Pollutant load characteristics are extracted based on the digital twin source-end model and the digital twin receiver-end model to obtain pollutant load characteristic data; The receiving end pollution load data is obtained by performing pollution load characteristic data processing on the receiving end. Based on the pollution carrying data at the receiving end, pollution migration simulation is performed on the digital twin source model and the digital twin receiver model to obtain pollution migration data; Based on pollution migration data, the source-end model and receiver-end model of the digital twin are matched and labeled with pollution solid waste to obtain the second solid waste matching model. The extraction of pollutant load characteristics includes: Pollutant data is obtained by extracting pollutant data based on the digital twin source-end model; Interaction path data is obtained by extracting interaction paths based on the source model and receiver model of the digital twin. Pollution potential attenuation processing is performed on the interaction path data based on the pollutant data to obtain pollutant path data; Multi-step particle migration simulation is performed on pollutant path data and preset electronic map data to obtain multi-step particle migration data. Path overlap diffusion enhancement was performed based on multi-step particle migration data to obtain pollution exposure map data; Based on the pollution exposure map data, a pollution migration field-receiver load simulation was performed on the digital twin receiver model to obtain pollutant load characteristic data.

5. The method according to claim 1, characterized in that, The path risk avoidance and control includes: Path risk extraction was performed on the first solid waste matching model and the second solid waste matching model to obtain the first path risk data and the second path risk data, respectively. A path risk map is constructed based on the first path risk data and the second path risk data to obtain path risk map data. By performing avoidance processing on the path risk map data, a solid waste matching model is obtained.

6. The method according to claim 1, characterized in that, S4 include: Task scheduling units are extracted based on the solid waste matching model to obtain task scheduling unit data. The task scheduling unit data is subjected to task splitting criteria to obtain task splitting data; Risk-reorganized task split data is performed to obtain task reorganized data; Feasibility screening of task reorganization data was performed to obtain solid waste scheduling data.

7. A digital twin-driven real-time matching and optimization system for solid waste, characterized in that, For executing the digital twin-driven real-time solid waste matching and optimization method as described in claim 1, the digital twin-driven real-time solid waste matching and optimization system comprises: The source-end twin modeling module is used to acquire source-end data of construction solid waste and perform source-end twin modeling of solid waste based on the source-end data to obtain a digital twin source-end model. The receiver twin modeling module is used to acquire data from the construction solid waste receiver and perform twin modeling of the receiving point based on the data to obtain a digital twin receiver model. The solid waste matching and modeling module is used to match the digital twin source model and the digital twin receiver model to obtain a solid waste matching model. The task splitting and fusion scheduling module is used to perform task splitting and fusion scheduling based on the solid waste matching model to obtain solid waste scheduling data.

Images