Urban garbage clearance scheduling method and system based on path optimization

Abstract

The invention discloses an urban garbage collection and transportation scheduling method and system based on path optimization, and relates to the technical field of urban garbage collection and transportation and intelligent scheduling optimization, and the method comprises the steps: collecting multi-source original data, carrying out the standardization and time-space registration, and generating the structured observation data of vehicles, roads and collection and transportation points; constructing a dynamic garbage quantity prediction model according to the structured observation data, and predicting garbage increment and liquid level trend of each collection and transportation point; generating a task package and establishing a mapping relation based on a dynamic garbage quantity prediction model result in combination with road passage, vehicle load and time window constraints; and constructing a multi-target path optimization model on the basis of the mapping relation and the constraint, outputting an optimal driving path of the vehicle, and converting a path optimization result into a scheduling instruction. According to the method, the task package is generated on the basis of the dynamic garbage quantity prediction result in combination with road traffic, vehicle load and time window constraints, hierarchical matching of task targets and vehicle resources is achieved, and therefore the reasonability of task allocation and operation feasibility are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of urban garbage collection and intelligent scheduling optimization, in particular to a garbage collection and scheduling method and system based on path optimization. BACKGROUND

[0002] With the continuous acceleration of urbanization, the amount of household garbage continues to grow, and the urban collection system gradually shifts from manual experience scheduling to automatic scheduling based on information technology and intelligence. Existing garbage collection systems rely on the Internet of Things, geographic information systems (GIS) and vehicle terminals to achieve data collection and positioning monitoring, and complete the collection operation through static path planning or periodic route setting. However, the generation of urban garbage is affected by many factors such as population density, climate change, regional function and time period rules, and its spatio-temporal distribution has significant dynamic and non-equilibrium characteristics. Traditional collection and scheduling methods based on fixed routes or single indicators cannot achieve multi-objective coordination and optimization, and cannot meet the requirements of modern cities for collection efficiency, energy balance and real-time response.

[0003] The existing urban garbage collection and scheduling technology mainly has the following problems: first, there is a lack of unified structured processing mechanism at the data level. The current collection system collects liquid level, weighing, road and vehicle operation data from scattered sources with varying degrees of precision, lacks a unified standardization and spatio-temporal registration process, making it difficult for prediction models and scheduling algorithms to operate under the same data benchmark, and there is a lag and deviation in model input. Secondly, the prediction link fails to achieve dynamic quantification and spatial correlation. Most collection methods only rely on historical averages or manual experience to set collection frequencies, without establishing a dynamic prediction model that can simultaneously depict time series and spatial coupling relationships, making it impossible to accurately identify the garbage accumulation patterns of different points at different times, resulting in a disconnect between collection timing and vehicle scheduling. Thirdly, the task scheduling process lacks a path optimization mechanism that integrates multiple constraints. Traditional algorithms usually take the shortest path or minimum travel as a single objective, without considering actual constraint factors such as vehicle load, road traffic level, operation time window and risk priority, resulting in low vehicle utilization, high travel energy consumption and increased risk of local area overflow. In addition, existing collection systems are mostly limited to static path planning, without establishing a closed-loop logic from data collection, prediction modeling, task construction to path optimization, making it impossible to update the model adaptively and correct the scheduling dynamically during operation. In summary, the existing technology has obvious shortcomings in data fusion, dynamic prediction, multi-objective path optimization and execution feedback, making it difficult to achieve intelligent scheduling and fine management of urban garbage collection throughout the entire process. SUMMARY

[0004] In view of the above problems, the present application is proposed.

[0005] Therefore, the technical problem solved by the present application is that the existing urban garbage collection and transportation scheduling method has problems such as insufficient data standardization, low dynamic prediction accuracy, single path optimization, lack of model adaptive update, and how to realize dynamic prediction, constraint fusion and multi-objective path optimization scheduling of urban garbage collection and transportation tasks based on multi-source perception data.

[0006] To solve the above technical problems, the present application provides the following technical solutions: a city garbage collection and transportation scheduling method based on path optimization, including collecting multi-source original data for standardization and space-time registration, generating structured observation data of vehicles, roads and collection and transportation points; constructing a dynamic garbage quantity prediction model according to the structured observation data, predicting the garbage increment and liquid level trend of each collection and transportation point; generating a task package based on the results of the dynamic garbage quantity prediction model combined with road traffic, vehicle load and time window constraints and establishing a mapping relationship; constructing a multi-objective path optimization model based on the mapping relationship and constraints, outputting the optimal driving path of the vehicle, and converting the path optimization result into a scheduling instruction.

[0007] As a preferred scheme of the city garbage collection and transportation scheduling method based on path optimization, the multi-source original data includes liquid level sensor signals, weighing data, road state information, vehicle operation information and operation rule data.

[0008] As a preferred scheme of the city garbage collection and transportation scheduling method based on path optimization, the standardization and space-time registration includes unit unification, time stamp synchronization, geographic coding mapping and abnormal data elimination processing of the collected multi-source original data.

[0009] As a preferred scheme of the city garbage collection and transportation scheduling method based on path optimization, the dynamic garbage quantity prediction model adopts a time series and space correlation joint modeling method, integrates historical liquid level, collection frequency and adjacent point change characteristics, generates garbage increment and liquid level proportion prediction results for the next period, and outputs a structured data table containing point number, predicted time and task priority.

[0010] As a preferred scheme of the city garbage collection and transportation scheduling method based on path optimization, the generation of the task package includes constructing vehicle load constraints, operation time window constraints and road traffic constraints based on the results of the dynamic garbage quantity prediction model, grouping the task points using a spatial clustering algorithm, generating a task package structure body with block identification, and establishing a mapping relationship table between the task points, vehicles and transfer stations.

[0011] As a preferred embodiment of the urban waste collection and dispatching method based on path optimization described in this invention, the multi-objective path optimization model integrates four objective items: travel distance, energy consumption, overflow risk, and time penalty. It sets capacity, time window, path continuity, and single service constraints, solves the optimal path scheme through mixed integer linear programming, and generates a vehicle task execution order and trip data table.

[0012] As a preferred embodiment of the urban waste collection and dispatching method based on route optimization described in this invention, the route optimization results include vehicle number, task sequence, estimated arrival time, loading capacity and transfer station information. The route optimization results are converted into a dispatching instruction set and sent to the vehicle terminal. At the same time, a bidirectional index relationship between task number and vehicle number is established.

[0013] Another objective of this invention is to provide a path optimization-based urban waste collection and dispatching system that can predict the increase in waste volume and liquid level trends at each collection point by constructing a dynamic waste volume prediction model based on structured observation data, thus solving the problem of low dynamic prediction accuracy in current urban waste collection and dispatching methods.

[0014] As a preferred embodiment of the urban waste collection and dispatching system based on path optimization described in this invention, it includes: a data acquisition and standardization module, a dynamic waste volume prediction module, a task package construction and constraint generation module, a multi-objective path optimization solution module, and a dispatching execution module; the data acquisition and standardization module is used to collect multi-source raw data, and perform spatiotemporal registration and structured observation data generation through unit unification, timestamp alignment, and geocoding; the dynamic waste volume prediction module is used to establish a dynamic prediction model based on the standardized data, and calculate the waste increment and liquid level change at each collection point by integrating time series and spatial correlation features. The system analyzes trends and generates prediction results and task priority datasets. The task package construction and constraint generation module constructs vehicle load, operation time windows, and road traffic constraints based on prediction results and standardized data. It uses spatial clustering to form task package structures and establishes a mapping relationship between task points, vehicles, and transfer stations. The multi-objective path optimization solution module establishes a multi-objective path optimization model under task mapping and constraint conditions. It solves the path by comprehensively considering multiple dimensions such as driving distance, energy consumption, risk, and time, outputting the optimal vehicle driving path and task execution order. The scheduling execution module converts the optimization results into vehicle scheduling instructions.

[0015] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a route-optimized urban waste collection and dispatching method.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a path-optimized urban waste collection and dispatching method.

[0017] The beneficial effects of this invention are as follows: The urban waste collection and dispatching method based on path optimization provided by this invention standardizes and spatiotemporally registers multi-source raw data, achieving unified encoding and structured processing of data from liquid level sensors, weighing equipment, road traffic, and vehicle operation. This establishes a spatiotemporally consistent observation data system, ensuring the consistency of input and data accuracy for subsequent model calculations. By constructing a dynamic waste volume prediction model based on structured observation data, it achieves time-series analysis and spatial correlation modeling of waste generation patterns at each collection point, enabling early detection of waste increments and liquid level trend changes, thus providing the dispatching system with a dynamic and predictable basis for waste collection demand. By generating task packages based on dynamic waste volume prediction results combined with road traffic, vehicle load, and time window constraints, and establishing mapping relationships, it achieves hierarchical matching of task objectives and vehicle resources. This maps complex waste collection tasks into executable vehicle dispatching units, thereby improving the rationality of task allocation and operational feasibility. By constructing a multi-objective path optimization model based on mapping relationships and constraints, this invention achieves joint solutions for multi-dimensional indicators such as distance, energy consumption, risk, and time. This model is used to generate optimal driving routes for each vehicle and form scheduling instructions, thereby achieving global coordination and optimal resource allocation for urban waste collection operations. Overall, this invention constructs a closed-loop system from data collection to path decision-making through the collaborative design of data standardization, dynamic prediction, constraint modeling, and multi-objective optimization, significantly improving the intelligence level of urban waste collection scheduling and the system's operational efficiency. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 The first embodiment of the present invention provides an overall flowchart of a route optimization-based urban waste collection and dispatching method. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0021] Example 1, referring to Figure 1 As an embodiment of the present invention, a method for urban waste collection and dispatching based on path optimization is provided, comprising: S1: Collect raw data from multiple sources, standardize and register them in time and space, and generate structured observation data of vehicles, roads and collection points.

[0022] Furthermore, multi-source raw data is collected from urban waste collection scenarios and registered categorized by type. The data collection targets include: collection points (such as community collection stations, small collection points, and commercial service locations), transfer stations, waste collection vehicles, road nodes and edges, and scheduling and traffic restriction rules. An access channel is established for each type of object. Sensor devices (level gauges, weight gauges, door sensors and operation buttons, vehicle weighing, OBD / motor controllers) are accessed through an industrial IoT gateway, using protocols such as Modbus-TCP / RTU, CAN, RS485, BLEMesh, or LoRaWAN, and uniformly converted to MQTT / HTTP streams at the edge. Information system data (traffic flow, construction site fencing, temporary activity applications, prohibited collection periods, administrative divisions, and community ledgers) are retrieved periodically via RESTful APIs or batch files. Vehicle location data is reported through GNSS terminals or mobile terminal SDKs and converted into a unified JSON structure by the access layer. All access channels are configured with token authentication, TLS transmission, source whitelist and frequency control threshold on the data entry side, and organized by topic or directory according to "object type / city / administrative region / device ID / date" for easy isolation and auditing.

[0023] It should be noted that a unified master data coding system is adopted. Collection points are assigned POINT_IDs, with the coding rule being city code + administrative region code + serial number; road nodes and edges are assigned NODE_ID / EDGE_IDs, and the ReferenceID of the external road network is retained; vehicles are assigned VEH_IDs, and static attributes such as license plate, vehicle type, energy type, and load capacity are recorded; transfer stations are assigned DEPOT_IDs, recording processing capacity, operating time windows, and the number of dumping sites; and data collection equipment is assigned DEV_IDs, recording firmware version, range, and calibration curve. Spatial location uses a unified coordinate reference, storing latitude, longitude, decimal precision, and elevation information; simultaneously, a GeoHash and grid code are generated for each spatial point for subsequent rapid neighborhood retrieval. Timestamps are uniformly stored using millisecond-level UNIX time, and dual timestamp fields (TS_SRC, TS_SRV) are recorded on both the reporting end and the server side for alignment.

[0024] Furthermore, NTP time synchronization is configured between the edge gateway and the central access layer. All reported data undergoes a time standardization process at the access layer: when the difference between TS_SRC and TS_SRV exceeds a threshold, CLOCK_DRIFT is marked and the correction amount is recorded; for batch files, the generation time in the filename or header is used as the base time, and if missing, the file landing time is used and marked TS_IMPUTED; GNSS trajectory data undergoes sequence integrity checks, and if reverse order or jumps are found, segmentation and correction are performed according to adjacent difference and velocity threshold rules. All time fields are uniformly converted to UTC for storage, while the local time zone offset field is retained for subsequent time window logic judgment.

[0025] It should be noted that the access layer performs format validation and mandatory field validation (ID, timestamp, coordinates, measured value). Unqualified records are entered into the exception channel and error codes are generated. Numerical measurements (liquid level, weight, vehicle load) undergo range validation and physical consistency validation; for example, liquid level percentage is limited to 0-100%, weight is limited to within the equipment's range, and vehicle load must not exceed the rated load. Trajectory data undergoes speed and acceleration threshold validation; points exceeding the threshold are smoothed using median filtering and sliding window mean. Duplicate reports are deduplicated (based on the same primary key and duplicate timestamps within the window). For breakpoints and missing segments, nearest-neighbor interpolation or step-hold strategies are used, and an IMPUTE_FLAG is added to the record. Pairing validation is performed on dumping and weighing records (entry time, exit time, net weight, train number); if inconsistent, it is marked as UNMATCHED_TICKET, awaiting subsequent manual or batch processing repair. Cleaning rules are stored in a configurable rule table, which supports version management and canary releases.

[0026] Furthermore, the system defines a data dictionary and a unit of measurement table to standardize units and scale common fields, such as weight to kilograms, volume to liters, distance to meters, and speed to m / s or km / h, while specifying decimal precision and rounding rules. Raw code values ​​reported by level and weight sensors from different manufacturers are converted into physical quantities based on equipment calibration curves. Energy and fuel consumption reports are uniformly converted to standard energy consumption values, retaining both original and converted values ​​in columns. Road levels and congestion levels reported by external traffic platforms are enumerated and mapped, standardized to four levels: FREE / FLOW / CONGESTION / LOCKED. Administrative division names are standardized to avoid aggregation errors caused by multiple spellings of the same name or historical names. The field standardization results are stored in columnar format in the ODS_standardization layer, and the source system, rule version, processing time, and processing node are recorded in the metadata.

[0027] It should be noted that an Object-Relationship-Attribute (ORA) ternary structure is constructed above the standardization layer: objects include collection points, road nodes, roadsides, vehicles, transfer stations, and operational rules; relationships include point-road connectivity, vehicle-affiliated parking lot, vehicle-bound equipment, point-administrative division, point-transfer station reachability, and point-classification attribute; attributes retain static attributes (such as load, classification type, and service time window) and dynamic attributes (such as current liquid level, predicted quantity placeholder field, and latest reporting time). Based on the ORA structure, a task candidate table is generated according to schedule or shift. The candidate table records the point ID, its region, waste category, allowed service time window, most recent service time, equipment availability, reachability flags to the nearest transfer station, and placeholder fields for subsequent priority and route optimization. The candidate generation process does not make scheduling decisions, but only ensures data integrity and availability.

[0028] Furthermore, the basic road network is imported, and topological verification and one-way / two-way attribute labeling are performed on nodes and edges, merging minor suspended edges and isolated nodes. A road network status overlay is generated based on temporary traffic organization, construction site fencing, and traffic restriction lists issued by administrative departments. The overlay includes an enumeration of drivable vehicle types, time-based traffic constraints, temporary no-entry signs, and an estimated travel time correction coefficient in the edge attributes. For each collection point, the nearest road network node connected to it is calculated (based on minimum Euclidean distance and pedestrian / operational road filtering), and the point is attached to the road network topology with the attachment error recorded. An reachability index table is output, containing the mapping between points and the road network, drivable signs for each vehicle type at different time periods, and a placeholder field for initial travel time.

[0029] It should be noted that the data landing adopts a layered design: Raw Data Layer (RAW) → Standardized Data Layer (ODS) → Thematic Wide Table (DWS). The RAW layer retains the original message and binary attachments; the ODS layer stores narrow tables after cleaning and unit standardization; the DWS layer constructs point wide tables, vehicle wide tables, road network edge wide tables, and task candidate wide tables for scheduling needs. To support efficient retrieval, composite indexes are established in the point, vehicle, and edge tables: using (city code, administrative division, GeoHash prefix, object type) as the composite index key; a clustered index (object ID, timestamp) is established for time-series data; and an R-Tree or H3 index is established to support spatial range queries. All layered tables are registered in the metadata service, generating data lineage, data quality indicators, and field-level comments for reference in subsequent steps.

[0030] Furthermore, after standardized modeling is completed, a unified data access interface and event flow are provided externally: batch access is provided through standardized query APIs / SQL views, and real-time subscription is provided through standardized event topics. Events include location status updates, vehicle status updates, road network status updates, and task candidate generation. The interface specification defines fields, types, units, required and optional fields, sample messages, and error codes in the open interface directory.

[0031] S2: Based on structured observation data, a dynamic waste volume prediction model is constructed to predict the waste increase and liquid level trend at each collection point.

[0032] Furthermore, based on the standardization layer and thematic wide table, the system calls upon basic fields from the task candidate table, including site identifier, current liquid level percentage, previous cycle waste increment, timestamp, spatial coordinates, container capacity, site category, and administrative division. To ensure temporal consistency of the prediction input, the system performs data time alignment according to a defined time standardization process, performing linear interpolation or nearest neighbor preservation on liquid level, weight, weather, and activity data at different sampling frequencies to ensure all input features correspond at a uniform granularity. The aligned time series samples are then sorted in ascending time order using site number plus time index as the primary key, generating the basic sample set for model input.

[0033] It should be noted that, based on the time series samples, exogenous factor fields are called, including: weather information (temperature, rainfall, wind speed, humidity), holiday indications, intensity of large-scale events, regional population flow increments, permitted operation periods, time intervals from the last cleanup to the present, and unique thermal codes for location categories.

[0034] Furthermore, a comprehensive prediction model based on time autoregression, spatial lag, and collection reset mechanism is used to predict the amount of waste in future time windows at each collection point, expressed as:

[0035] in, Indicates collection point At any moment The logarithmic form of the garbage accumulation index, Indicates collection point At any moment The logarithmic form of the garbage accumulation index, This represents the individual fixed effect at the collection and transportation point. Indicates the time lag factor. Indicates collection point Neighborhood set All adjacent points within the time frame at time 1 The average log-cumulative amount, The spatial lag weighting coefficient. Indicates collection point At any moment Vector of environmental and social influencing factors This represents the corresponding parameter vector. This indicates that the point is at time [time]. The status variable indicating whether the cleaning operation is complete (set to 1 if cleaning is complete, otherwise set to 0). This reflects the speed and intensity of the recovery trend of subsequent waste accumulation caused by the waste removal operation. This is a random disturbance term.

[0036] It should be noted that the prediction is expressed as:

[0037] in, For container capacity, To indicate the collection point At any moment The actual percentage of waste liquid level, This indicates the collection points calculated using a dynamic waste volume prediction model. At any moment The logarithmic form of the garbage accumulation index, Indicates collection point At any moment Predicted increase in waste, Indicates collection point At any moment The predicted percentage of waste liquid level.

[0038] Furthermore, parameters are estimated based on historical sample data for each location, using weighted least squares or ridge regression to estimate parameter sets in batches under location grouping (administrative district / category). The average travel time calculated from the road network status overlay layer, which is used for spatial weighting, is automatically normalized to the location category similarity. After training, the system generates a parameter configuration file and stores it in the model parameter library, retraining every 24 hours to ensure the prediction model is updated in real-time according to the city's status.

[0039] It should be noted that, in obtaining the predicted liquid level percentage Then, classification is performed based on the set overflow threshold (0.85). If within the prediction window... Memory exists If the record is not found, the point is marked as an emergency point; if the maximum predicted value is in the range of 0.65 to 0.85, it is marked as a deferred point. The identification results of emergency points and deferred points are associated with the fields of the task candidate table to generate a task list with priority flags (two categories: must-achieve tasks and deferred tasks).

[0040] Furthermore, a liquid level percentage threshold of 0.85 is used as the criterion for emergency point judgment. This threshold is not a fixed empirical value, but is determined comprehensively based on long-term collected data on physical capacity characteristics, statistical distribution of collection cycles, and container safety margins. Specifically, the data for each type of garbage bin is first extracted from the equipment master data table and weighing record table. In addition to historical full-load disposal records, an empirical distribution function for the liquid level percentage was calculated for each type of container. Statistical results show that when the liquid level percentage of most containers reaches the range of 0.80 to 0.90, the sensor liquid level signal and the weighing measurement value show a non-linear increase (due to compaction or accumulation effect), and at this time, the resident disposal outlet still has a physical buffer space of 10% to 20%.

[0041] To balance overflow risk and scheduling margin, a safety margin factor is set. The liquid level threshold is expressed as:

[0042] in, Indicates the liquid level threshold. This represents the average liquid level distribution (approximately 0.78). The standard deviation (approximately 0.07) represents the safety margin factor. Take 1. Substituting, we get This threshold ensures that the waste container still has a short-term safe space when the liquid level exceeds 85%, and also triggers the collection task one scheduling cycle in advance to prevent the liquid level dead zone from occurring after the sensor saturates. Therefore, 0.85 is not an arbitrary constant, but an empirically stable value derived from the physical capacity of the container, the measurement nonlinearity range, the historical collection time interval distribution, and safety margin parameters.

[0043] It should be noted that the system allows the threshold to be adjusted parametrically for different container categories (kitchen waste, recyclables, other waste) or seasonal changes: for example, it can be 0.80 during the high temperatures of summer, and 0.90 during winter or in low-frequency disposal areas; these adjustments are maintained in the configuration file in the form of a threshold mapping table and recorded in the version information of the metadata service to ensure that the system has consistent traceability and adjustability under different operating environments.

[0044] S3: Based on the results of the dynamic waste volume prediction model, combined with road traffic, vehicle load and time window constraints, generate task packages and establish mapping relationships.

[0045] Furthermore, based on the generated wide table of candidate tasks, the system reads relevant field information for each collection point, including point identifier, administrative division, associated transfer station number, current liquid level percentage, predicted liquid level percentage, predicted waste increment, container capacity, waste category code, and task priority label (urgent task vs. deferred task). The system continues to use a ternary modeling approach, describing the global constraint structure of the urban waste collection network through the relationships between point objects, road node objects, vehicle objects, transfer station objects, and operational rule objects. During data retrieval, the system uses the standardized field mapping table defined in the first step to ensure consistency in units such as capacity, time, weight, and liquid level, and to unify the timestamp format and decimal precision.

[0046] It should be noted that, along with the imported task data, the generated road network status overlay and accessibility index table are loaded. This table includes the node number, road grade, traffic direction, traffic restrictions, construction status, and permitted vehicle types for each road. Based on this information, the system establishes a road accessibility judgment mechanism: when a road is passable in the current time period and the vehicle type meets the road grade requirements, the road is marked as an accessible route; if the road is under construction, subject to traffic restrictions, or exceeds the vehicle's tonnage class, it is marked as an impassable route. Vehicle data comes from the vehicle master data table, including fields such as vehicle number, vehicle type, rated load, energy type, and affiliated transfer station. By comparing vehicle type and road grade, a vehicle-road matching table is established to clarify the set of roads each vehicle can travel on in different areas.

[0047] Furthermore, a dual constraint of capacity and time window is established for each task location. The capacity constraint is jointly limited by the container's rated capacity and the vehicle's approved load capacity, ensuring that the total amount of waste loaded by a single vehicle during a collection cycle does not exceed the vehicle's maximum carrying capacity. The time window constraint is defined by an administrative division operation rule table, which includes the allowed collection time periods, noise limit levels, and holiday adjustment information for each region. All time information is stored in a unified Coordinated Universal Time (UTC) format with an attached time zone offset field to ensure consistent time calculations during cross-regional scheduling. During task generation, if the estimated operation time of a task is outside the allowed time window, the task is automatically marked as postponed to avoid violating regional operation regulations. All constraint parameters are stored in configuration tables and can be adjusted independently by region, waste type, or vehicle category during system operation.

[0048] It should be noted that after completing the basic constraint screening, the remaining task points are clustered. The clustering process first spatially groups adjacent points based on the generated geocoding prefix, ensuring that points within the same group have close geographical adjacency. Then, the system combines road accessibility data to further check whether there are continuous passable paths between points in the same group. If they are not connected, they are automatically split into independent groups. Based on spatial clustering, the system calculates the average predicted liquid level of points in each group and counts the proportion of emergency tasks. When the proportion of emergency tasks exceeds a set threshold (default value is 40%), the group is marked as a priority collection area. Subsequently, the system performs a secondary screening of task points in each area based on waste category, operation time window, and administrative division, grouping points with the same category, similar time, and spatial connectivity into a task package. Each task package contains a unique task number, a set of task points, an administrative division code, waste category, a unified time window, an average liquid level percentage, and a corresponding transfer station number. All task package data is written to a new task package table and automatically associated with vehicle and area information during generation.

[0049] Furthermore, after the task package is generated, its internal consistency is verified by the system. First, a capacity consistency check is performed: the total predicted waste volume of all points within the task package is calculated. If the result exceeds 80% of the vehicle's rated load capacity, the system automatically splits the task package into two or more sub-packages. Second, a time window consistency check is performed: if the overlapping service time intervals of each point within the task package are less than the system's minimum allowed duration (default 15 minutes), the task package is split into task sets for different time periods. Road connectivity is checked using the reachability index table, calculating the shortest path between any two points within the task package; if the path passes through temporarily restricted roads or crosses construction areas, the task package is marked as infeasible and enters the adjustment queue. Finally, the system also performs a waste category compatibility check to ensure that the same task package contains only waste points of the same type; points of different categories will be forcibly assigned to different task packages. Through these steps, a three-way mapping table between tasks, vehicles, and transfer stations is generated to record the candidate vehicle and transfer station numbers corresponding to each task package, providing complete input parameters for the next step of route optimization.

[0050] It should be noted that, based on the established transit station master data table and its point-transit station reachability relationships in the ORA structure, the most suitable transit station node is determined for each task package. The selection logic primarily considers the shortest average travel time and the current load of the transit station, prioritizing the node with the shortest travel time and lowest load among multiple reachable transit station candidates. For task packages crossing administrative regions, a cross-regional flag is automatically added to the administrative region mapping table to prompt the subsequent scheduling module to perform cross-regional approval and permit verification during route planning. This cross-regional flag is retained in the task package field during system export to ensure that subsequent steps can accurately identify cross-regional tasks.

[0051] Furthermore, all task package data that has undergone consistency verification and transfer station mapping updates is uniformly written into the task package table of the topic wide table, and provided to the path optimization module through the interface service. Output fields include task number, region code, waste category, feasible vehicle set, time window range, maximum carrying capacity, transfer station number, average predicted liquid level percentage, and reachable path set, etc.

[0052] S4: Construct a multi-objective path optimization model based on mapping relationships and constraints, output the optimal driving path for vehicles, and transform the path optimization results into scheduling instructions.

[0053] Furthermore, all input fields required for path optimization are read, including task number, task point set, area code, predicted liquid level percentage, predicted waste increment, current liquid level percentage, container capacity, vehicle number, vehicle rated load, unit energy consumption coefficient, road distance, operation time window, and set of passable paths. All fields undergo unit conversion in the standardization layer (distance in meters, time in milliseconds, load in kilograms), and include the time zone offset field TIMEZONE_OFFSET. A node reachability matrix is ​​constructed through a road network state overlay layer to determine the shortest path structure between task points and transfer stations, and a set of feasible vehicle paths is generated at the task package level as the input domain of the optimization model.

[0054] It should be noted that the multi-objective optimization model, which includes distance, energy consumption, risk, and time penalty, is expressed as follows:

[0055] in, Indicates the overall optimization objective. This represents the target weight parameter, which comes from the scheduling parameter table, and the default ratio is 4:3:2:1.

[0056] Distance to target Represented as:

[0057] in, Indicates collection point to The road distance, Indicates vehicle From the collection point Drive to binary variables, Indicates a collection of vehicles. This indicates a set of objects along the roadside.

[0058] Energy consumption and carbon emission targets Represented as:

[0059] in, This represents the vehicle's energy consumption coefficient per unit. This represents the load correction factor. Indicates vehicle At the collection point The load capacity, This indicates the vehicle's approved load capacity.

[0060] Spillover risk target item Represented as:

[0061] in, Indicates the collection point set. Indicates the spillover risk weight. Indicates collection point Whether it is overwritten by the currently scheduled task package. If it is overwritten, the value is 1; otherwise, the value is 0.

[0062] Time penalty target item express:

[0063] in, This represents the default amount within the time window. If the vehicle arrives later than the upper limit, the difference is taken.

[0064] Four types of core constraints are set: capacity constraints, time window constraints, path continuity constraints, and single service constraints.

[0065] ① Capacity constraint: The total load of each vehicle during the scheduling cycle shall not exceed the approved load capacity.

[0066] ②Time window constraint: The time it takes for the vehicle to arrive at the mission point must fall within the allowed time interval.

[0067] ③ Path continuity constraint: Each vehicle must depart from its respective transfer station and eventually return.

[0068] ④ Single service constraint: Each task point can only be served by one vehicle within a scheduling cycle.

[0069] After the model is built, a mixed-integer linear programming algorithm is used to solve for the optimal scheduling scheme. During the solution process, the system normalizes the four objective terms to eliminate calculation biases caused by different units.

[0070] The output includes the optimal driving route for each vehicle, the service order at each task point, the total driving distance, energy consumption, risk weighting, and time window deviation, and generates a path optimization result table. The fields in the result table correspond one-to-one with the input data, including vehicle number, task execution order, cumulative driving distance, total energy consumption, timeout, and risk indicators. All results are timestamped and include the model version number for subsequent scheduling execution and performance backtracking.

[0071] After the multi-objective path optimization model is solved, a vehicle dispatch instruction set is automatically generated. The instruction set uses the path optimization result table as its data source, and its core fields include vehicle number, task sequence, service point number, estimated arrival time, estimated load, predicted liquid level, task priority label, transfer station number, and return route information. This path data is distributed to vehicle terminals or mobile workstations in the form of task queues through the dispatch execution interface layer. Each task queue carries a unique task number, and a bidirectional index relationship is established in the database with the vehicle's primary key to ensure that the location information, weighing data, and task status generated by the vehicle during operation can be traced back to the specific model inputs and outputs.

[0072] During vehicle operation, the system continuously collects operational status information via the onboard terminal, including real-time vehicle location coordinates, operating speed, load weight, number of tipping operations, liquid level changes, and arrival and departure times. The collected data is uploaded to the real-time status acquisition layer via the IoT communication module, where it undergoes timestamp alignment and unit conversion according to standardized rules. The data field structure remains completely consistent with standard fields, such as latitude and longitude fields, timestamp fields, vehicle load, and task number. The system compares the vehicle's current location with the planned path nodes in the optimization model in real time. When the vehicle arrives at a point, the actual arrival time is matched with the planned arrival time, the actual deviation is calculated, and the TIME_OFFSET field in the execution monitoring table is updated. After each task is completed, the system uses the vehicle's returned weighing data, liquid level sensor feedback, and task completion flags to determine the completion status.

[0073] Example 2, an embodiment of the present invention, provides a method for urban waste collection and dispatching based on path optimization. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0074] First, the main urban area of ​​a city was selected as the experimental area, with a total of 60 garbage collection points, 4 transfer stations, and 10 garbage collection vehicles in 3 administrative districts within the urban area selected as the test subjects. The experimental period was set at 7 days, and the goal was to verify the performance improvement of the invention in scheduling after data standardization, dynamic prediction, and path optimization.

[0075] First, during the data acquisition phase, level gauges and weight sensors are installed at all collection points, and data is reported via an industrial IoT gateway using the LoRaWAN protocol. All data is uniformly converted to MQTT stream format and stored in the RAW raw layer through the access layer. The system performs format checks and anomaly removal on fields such as level, weight, timestamp, and latitude / longitude to ensure that the time error is less than ±2 seconds and the measurement data missing rate is less than 1%. Simultaneously, road network data and traffic restriction lists from the transportation department are imported to form a road network status overlay layer. Each waste collection vehicle is equipped with a GNSS positioning terminal and an onboard weighing system, and vehicle data is uploaded to the ODS standardization layer in real time. All numerical data uses a unified unit: weight in kilograms, volume in liters, and distance in meters.

[0076] Secondly, after data standardization, a structured observation dataset was constructed. A unique code POINT_ID was assigned to each collection point, and a VEH_ID was assigned to each vehicle. A relationship table was established linking vehicles, transfer stations, and task points. Through spatiotemporal registration, the coordinates of collection points, road nodes, and transfer stations were unified to the WGS-84 coordinate system, and GeoHash codes were generated for subsequent spatial clustering. Timestamp fields were uniformly converted to UTC milliseconds, and local timezone offset values ​​were recorded.

[0077] Subsequently, based on standardized data, a thematic wide table was used to extract information on liquid level, weight, weather, and holidays, which was then input into the dynamic waste volume prediction model. The system was trained on historical samples from 60 collection points using a 3-week sliding window, outputting the waste increase and liquid level trends for the next 24 hours. Emergency points were automatically marked by setting a liquid level threshold of 0.85, generating a task list. Results showed that approximately 18 points had predicted liquid levels above 0.85 within the next day's time window and were prioritized for mandatory collection by the system.

[0078] Next, based on the generated prediction results, the system calls upon data such as road traffic, vehicle load, and operation time windows to construct capacity and time constraints. A spatial clustering algorithm is used to aggregate geographically adjacent points with overlapping time windows into task packages, forming 12 executable task groups. Each task group establishes a mapping between a task package table and a vehicle matching table, ensuring that the load of a single task does not exceed 85% of the vehicle's rated load.

[0079] Finally, the system executes a multi-objective path optimization model under task mapping and constraints. The model completes the solution within 120 seconds, outputting the optimal driving path, task point order, and estimated arrival time for each vehicle. The optimization results are converted into scheduling instructions and sent to the vehicle terminals. During execution, the vehicles transmit location and weighing data in real time, and the system automatically records the actual arrival time and deviation. Experimental data is continuously collected for one week, ultimately generating a path optimization result table and execution log data for performance comparison and analysis.

[0080] Table 1 Experimental Data

[0081] As shown in Table 1, after implementing the method of this invention, the driving distance, energy consumption, task completion time, and overflow rate of each vehicle remained within a relatively optimal range. Compared with historical data using the traditional static scheduling method (average driving distance of approximately 50–55 km, energy consumption of approximately 18.5 L / 100 km, and overflow rate of approximately 3%), the system achieved significant improvements through multi-source data standardization and path optimization models.

[0082] First, during the data acquisition and standardization phase, a unified data coding system and spatiotemporal registration were used to avoid time drift issues caused by asynchronous reporting from multiple sensors. The reporting time error of all devices was controlled within ±2 seconds, making the input data for the prediction model more consistent and reducing noise propagation. This accuracy provides an important foundation for the predictive reliability of subsequent models.

[0083] Secondly, in the dynamic waste volume prediction stage, the system, based on time series and spatial correlation analysis, can accurately identify high-risk locations. In the experiment, 17 out of 18 locations with predicted liquid levels exceeding 0.85 actually showed near-overflowing waste, achieving a prediction accuracy of 94.4%, while traditional methods based on historical averages only achieved about 68%. This indicates that the model of this invention can effectively reflect the spatiotemporal dynamic characteristics of waste generation, triggering collection tasks in advance and reducing the risk of overflow.

[0084] Third, during the task package construction phase, the system integrates vehicle load capacity and operation time window constraints, automatically aggregating tasks of the same type and location within the same area, reducing redundant scheduling and empty runs. In the experiment, the average load utilization rate of each vehicle reached 83%, significantly higher than the 72% level of traditional manual scheduling, indicating that this method can achieve balanced task allocation under capacity constraints.

[0085] Fourth, during the route optimization phase, the system simultaneously considers four indicators—distance, energy consumption, time, and risk—through a multi-objective trade-off mechanism. After optimization, the average travel distance decreased by approximately 18%, energy consumption decreased by approximately 12%, and the average task completion time was shortened by 8–10 minutes. The overflow task rate was controlled at 0.7–0.9%, significantly better than traditional scheduling methods (approximately 3%). This indicates that this method can effectively coordinate vehicle travel and risk priorities, improving the overall scheduling effectiveness.

[0086] Finally, during the execution feedback phase, the real-time data reported by the vehicles and the optimized path form a closed loop, allowing the model parameters to be dynamically corrected. In the experiment, the system automatically detected the average path deviation at 3.5 minutes and could adaptively adjust the time window parameters and energy consumption weights in the next cycle to achieve continuous optimization.

[0087] In summary, through standardized multi-source data acquisition, dynamic prediction, constraint modeling, and collaborative design of path optimization, this invention achieved a scheduling effect of "short distance, high load, low energy consumption, and low risk" in experiments. Compared with existing static planning methods, this invention not only significantly improves data consistency and prediction accuracy, but also demonstrates innovative advantages in path optimization and energy economy during scheduling execution, verifying the engineering feasibility and creative value of this method in intelligent urban waste disposal systems.

[0088] Example 3, an embodiment of the present invention, provides an urban waste collection and dispatching system based on path optimization, including a data acquisition and standardization module, a dynamic waste volume prediction module, a task package construction and constraint generation module, a multi-objective path optimization solution module, and a dispatching execution module.

[0089] The data acquisition and standardization module collects raw data from multiple sources, performs spatiotemporal registration and generates structured observation data through unit unification, timestamp alignment, and geocoding. The dynamic waste volume prediction module builds a dynamic prediction model based on standardized data, integrates time series and spatial correlation features, calculates the waste increment and liquid level change trends at each collection point, and generates prediction results and task priority datasets. The task package construction and constraint generation module constructs vehicle load, operation time window, and road traffic constraints based on prediction results and standardized data, uses spatial clustering to form a task package structure, and establishes a mapping relationship between task points, vehicles, and transfer stations. The multi-objective path optimization solution module builds a multi-objective path optimization model under task mapping and constraint conditions, solves the path by integrating multi-dimensional objectives such as travel distance, energy consumption, risk, and time, and outputs the optimal vehicle travel path and task execution order. The scheduling execution module converts the optimization results into vehicle scheduling instructions.

[0090] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0091] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0092] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0093] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0094] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for urban waste collection and dispatching based on path optimization, characterized in that, include: Collect raw data from multiple sources, standardize and register them in time and space, and generate structured observation data of vehicles, roads and collection points; A dynamic waste volume prediction model was constructed based on structured observation data to predict the increase in waste volume and liquid level trends at each collection and transportation point. Based on the results of the dynamic waste volume prediction model, combined with road traffic, vehicle load and time window constraints, task packages are generated and mapping relationships are established. A multi-objective path optimization model is constructed based on mapping relationships and constraints, outputting the optimal driving path for vehicles, and the path optimization results are transformed into scheduling instructions.

2. The urban waste collection and dispatching method based on path optimization as described in claim 1, characterized in that: The multi-source raw data includes liquid level sensor signals, weighing data, road condition information, vehicle operation information, and operation rule data.

3. The urban waste collection and dispatching method based on path optimization as described in claim 2, characterized in that: The standardization and spatiotemporal registration process includes unifying the units of the collected multi-source raw data, synchronizing the timestamps, mapping the geocoding, and removing abnormal data.

4. The urban waste collection and dispatching method based on path optimization as described in claim 3, characterized in that: The dynamic waste volume prediction model adopts a joint modeling approach of time series and spatial correlation, which integrates historical liquid level, collection frequency and changes in adjacent locations to generate prediction results of waste increment and liquid level ratio for the next period, and outputs a structured data table containing location number, prediction time and task priority.

5. The urban waste collection and dispatching method based on path optimization as described in claim 4, characterized in that: The generated task package includes constructing vehicle load constraints, operation time window constraints, and road traffic constraints based on the results of the dynamic waste volume prediction model, grouping task points using a spatial clustering algorithm, generating a task package structure with block identifiers, and establishing a mapping relationship table between task points, vehicles, and transfer stations.

6. The urban waste collection and dispatching method based on path optimization as described in claim 5, characterized in that: The multi-objective path optimization model integrates four objectives: travel distance, energy consumption, overflow risk, and time penalty. It sets constraints on capacity, time window, path continuity, and single service, and solves the optimal path scheme through mixed integer linear programming, generating a table of vehicle task execution order and trip data.

7. The urban waste collection and dispatching method based on path optimization as described in claim 6, characterized in that: The route optimization results include vehicle number, task sequence, estimated arrival time, load capacity, and transfer station information. The route optimization results are converted into a scheduling instruction set and sent to the vehicle terminal. At the same time, a two-way index relationship between task number and vehicle number is established.

8. A system employing the urban waste collection and dispatching method based on path optimization as described in any one of claims 1 to 7, characterized in that: It includes a data acquisition and standardization module, a dynamic waste volume prediction module, a task package construction and constraint generation module, a multi-objective path optimization solution module, and a scheduling and execution module; The data acquisition and standardization module is used to collect raw data from multiple sources, and to perform spatiotemporal registration and generate structured observation data through unit unification, timestamp alignment and geocoding. The dynamic waste volume prediction module is used to establish a dynamic prediction model based on standardized data, integrate time series and spatial correlation features, calculate the waste increment and liquid level change trend at each collection point, and generate prediction results and task priority datasets. The task package construction and constraint generation module is used to construct vehicle load, operation time window and road traffic constraints based on prediction results and standardized data, form a task package structure using spatial clustering, and establish a mapping relationship between task points, vehicles and transfer stations. The multi-objective path optimization solution module is used to establish a multi-objective path optimization model under task mapping and constraint conditions, and solve the path by comprehensively considering multi-dimensional objectives such as driving distance, energy consumption, risk and time, and output the optimal vehicle driving path and task execution order. The scheduling execution module is used to convert the optimization results into vehicle scheduling instructions.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the urban waste collection and dispatching method based on path optimization as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the urban waste collection and dispatching method based on path optimization as described in any one of claims 1 to 7.

Images