Intelligent scheduling system for garbage transfer station based on vehicle flow prediction

Abstract

The application relates to the technical field of urban solid waste management, and particularly discloses an intelligent scheduling system for a garbage transfer station based on traffic flow prediction, which comprises a data sensing module, a traffic flow prediction module, a fusion modeling module, a simulation analysis module and a collaborative scheduling module.The scheme realizes high-precision time-varying traffic flow prediction, establishes an accurate and quantitative dynamic mapping between scheduling instructions and future road conditions, searches for a shortest-time path in a time-varying road network, effectively avoids predictive congestion, predicts road section travel time, enables path planning decisions to respond to traffic state changes in a forward-looking manner, realizes automatic generation of a scheduling scheme, simulation verification, multi-objective optimization and standardized execution, generates a systematic initial solution for garbage point allocation, automatically generates a reasonable access order, and performs discrete-time simulation, so that the feasibility of an optimal scheme output in actual execution is ensured through dynamic optimization.

Description

Technical Field

[0001] This invention relates to the field of urban solid waste management technology, specifically to an intelligent scheduling system for garbage transfer stations based on traffic flow prediction. Background Technology

[0002] With the acceleration of urbanization and the continuous improvement of residents' living standards, the amount of urban domestic waste generated is showing a rapid year-on-year growth trend. As a key node in the urban solid waste collection and transportation system, waste transfer stations undertake the functions of concentrating, compressing, and transferring residential waste, and their operational efficiency directly affects the environmental sanitation quality of the entire city. Traditional vehicle dispatching schemes are usually based on static road networks or historical average road conditions for route planning. Route planning and dynamic traffic flow are heterogeneous, and dispatching decisions cannot perceive and avoid traffic congestion in future periods, resulting in frequent traffic congestion for collection vehicles. Moreover, traditional dispatching schemes have experience differences when there are many waste collection points and many vehicles, resulting in unstable dispatching quality, lack of simulation verification and multi-dimensional constraints, and difficulty in finding a globally optimal solution. Summary of the Invention

[0003] To address the above issues and overcome the shortcomings of existing technologies, this invention provides an intelligent scheduling system for garbage transfer stations based on traffic flow prediction. Addressing the heterogeneity between path planning and dynamic traffic flow in traditional scheduling schemes, this solution achieves high-precision time-varying traffic flow prediction through the iTransformer model and directly embeds the traffic flow speed matrix into the time-varying path modeling unit. This establishes a precise and quantifiable dynamic mapping between scheduling instructions and future road conditions. An improved Dijkstra algorithm is used to search for the shortest time path in the time-varying road network, with edge weights dynamically changing according to the time of vehicle entry into the road segment, effectively avoiding predictive congestion. The BPR function is used to predict the travel time of the road segment, making the path... Planning decisions can proactively respond to changes in traffic conditions, rather than passively adapting to existing congestion. Addressing the issue of unstable scheduling quality and difficulty in finding globally optimal solutions in traditional scheduling schemes, this solution automates the generation, simulation verification, multi-objective optimization, and standardized execution of scheduling schemes. It uses the K-means clustering algorithm to generate systematic initial solutions for waste point allocation, solves the traveling salesman problem within each vehicle's task subset to automatically generate a reasonable access order, and performs discrete-time simulations on each candidate scheduling scheme. Dynamic optimization ensures the feasibility of the final optimal solution in actual implementation, significantly improving the intelligence level and operational efficiency of the urban waste collection and transportation system.

[0004] The intelligent scheduling system for garbage transfer stations based on traffic flow prediction provided by this invention includes a data sensing module, a traffic flow prediction module, a fusion modeling module, a simulation analysis module, and a collaborative scheduling module.

[0005] The data perception module collects traffic flow data, operation and energy consumption data of collection vehicles, and operation data of garbage transfer stations and garbage collection points in residential areas within the collection responsibility area. The above data are sent to the traffic flow prediction module, the fusion modeling module and the simulation analysis module respectively.

[0006] The traffic flow prediction module builds a traffic flow prediction model based on the iTransformer architecture, uses historical traffic flow data to predict the traffic flow speed of each road segment on the road network in the future time period, outputs a traffic flow speed matrix and sends it to the fusion modeling module.

[0007] The fusion modeling module constructs a time-varying path model, a vehicle instantaneous energy consumption model, a time-varying queuing model, and a transfer service model based on the traffic flow velocity matrix, and couples them into a fusion model which is then sent to the simulation analysis module.

[0008] The simulation analysis module constructs a set of candidate scheduling schemes, uses a fusion model to simulate and extrapolate different candidate scheduling schemes, predicts multi-dimensional evaluation indicators, and generates a complete simulation evaluation report.

[0009] The collaborative scheduling module uses an improved genetic algorithm based on the simulation evaluation report to solve the optimal scheduling scheme for collection and transportation vehicles, and then distributes the solution to the on-board terminals of each collection and transportation vehicle and the scheduling control console of the transfer station.

[0010] Furthermore, the traffic flow prediction module includes a data preprocessing unit, a spatiotemporal feature extraction unit, and a model prediction unit;

[0011] The data preprocessing unit performs missing value imputation, outlier removal, and normalization on historical traffic flow data, and sets a fixed time step to aggregate the normalized data into a time series matrix.

[0012] The spatiotemporal feature extraction unit models the road network as a graph, using each road segment as a node and the spatial connections between road segments as edges to construct a graph structure. Based on the time series matrix, it calculates the weights between adjacent nodes through a graph attention mechanism, aggregates the traffic state information of adjacent road segments to the current node, and extracts spatiotemporal features that fuse local spatial correlations. The formula used is as follows: ; ;

[0013] In the formula, Represents a road network map. This represents a set of nodes that includes all road segments in the road network. This represents an edge set that contains the spatial connections between all road segments. and Indicates the index of the node. Represents a node The spatiotemporal characteristics, Represents a node The set of neighboring nodes, Represents a node For nodes Attention coefficient This represents the learnable weight matrix. Represents a node The time series matrix, Indicates the activation function;

[0014] The model prediction unit constructs a traffic flow prediction model based on the iTransformer architecture. It inputs the spatiotemporal features of each road segment into the Transformer encoder, calculates the spatial correlation between different road segments through a self-attention mechanism, predicts the traffic flow speed of each road segment in the future time period, and aggregates them into a traffic flow speed matrix for output. The formula used is as follows: ;

[0015] In the formula, This represents a matrix representing the predicted traffic flow speeds over a future time period. Represents a set of spatiotemporal feature sequences within a historical time window. This represents the nonlinear mapping function of the iTransformer model.

[0016] Furthermore, the fusion modeling module includes a travel time calculation unit, a time-varying path modeling unit, a vehicle energy consumption modeling unit, a charging queue modeling unit, a transfer service modeling unit, and a model fusion unit;

[0017] The fusion modeling module includes a travel time calculation unit, a time-varying path modeling unit, a vehicle energy consumption modeling unit, a charging queue modeling unit, a transfer service modeling unit, and a model fusion unit.

[0018] The travel time calculation unit fits the predicted traffic flow of each road segment to the traffic flow velocity matrix, and uses the BPR road resistance function to calculate the predicted travel time of collection vehicles through each road segment in each time period. The formula used is as follows: ;

[0019] In the formula, Indicates the road segment index. This indicates the time period index for when collection vehicles enter this road segment. Indicates the time period for collection vehicles Passing section Predicted travel time Indicates road segment Travel time of collection vehicles when there is no congestion. Indicates time period Time Section Predicted traffic flow Indicates road segment Road capacity and The parameters to be determined for the BPR road resistance function are 0.15 and 4.0, respectively.

[0020] The time-varying path modeling unit constructs a time-varying road network model based on the time when the collection vehicles enter the road segment and the predicted travel time in each time period. It uses an improved Dijkstra algorithm to search for the shortest time path from the starting point to the ending point in the time-varying road network model, calculates the earliest arrival time of each node, expands the nodes in the order of the earliest arrival times, and outputs the optimal path and its total travel time.

[0021] The vehicle energy consumption modeling unit establishes an instantaneous energy consumption model for the collection and transportation vehicles based on their load status and driving speed. The formula used is as follows: ;

[0022] In the formula, Indicates that the vehicle is on the road section Upper session Energy consumption during driving This indicates the base energy consumption rate when the vehicle is unloaded. Indicates the load-bearing energy consumption coefficient. Indicates the current load of the vehicle. The energy consumption coefficient represents the square of the velocity. Indicates road segment During the period Predicted traffic speed;

[0023] The charging queuing modeling unit establishes a time-varying queuing model based on the service capacity of charging stations. It fits the daily driving distance and travel probability distribution of the vehicles based on historical data on vehicle operation and energy consumption, and calculates the arrival rate of vehicles at charging stations using the following formula: ; ; ;

[0024] In the formula, Indicates a charging station. Indicates charging station During the period The arrival rate of collection and transportation vehicles This indicates the penetration rate of collection and transportation vehicles. Indicates the time from the start of the trip to the specified moment. Drive to charging station The cumulative driving distance, Indicates the maximum daily driving distance. Let be the probability density function of the daily travel distance. This represents the average queue length. This represents the average waiting time in the queue. Indicates charging station The number of charging stations Indicates the service rate per pile. Indicates the vehicle's battery capacity. This represents the target state of charge, set to 0.8. Indicates the state of charge upon arrival. Indicates charging power;

[0025] The transfer service modeling unit calculates the average waiting time at the waste transfer station for each time period and establishes a transfer service model. The formula used is as follows: ; ;

[0026] In the formula, Indicates a transit station. This represents the average waiting time. Indicates transit station During the period The arrival rate of collection and transportation vehicles This indicates the service rate per unloading station. This indicates the number of unloading positions at the transfer station. This indicates the busy rate of the transfer station service. Indicates the summation index;

[0027] The model fusion unit couples the time-varying path model, vehicle instantaneous energy consumption model, time-varying queuing model, and transfer service model into a fusion model.

[0028] Furthermore, the simulation analysis module includes a scheduling scheme generation unit, a simulation deduction unit, an index calculation unit, and an evaluation report generation unit;

[0029] The scheduling scheme generation unit uses the K-means clustering algorithm to divide the garbage points into groups based on the operation data, time window constraints, and number of available vehicles of each residential garbage point within the collection and transportation responsibility area. Each group corresponds to the task range of a collection and transportation vehicle. The traveling salesman problem is solved in the task subset of each collection and transportation vehicle to generate a preliminary set of candidate scheduling schemes. Each candidate scheduling scheme is a structured instruction set that specifies the garbage point access sequence, charging strategy, and transfer station arrival time for each collection and transportation vehicle.

[0030] The simulation and deduction unit performs discrete-time simulation for each candidate scheduling scheme, sets the simulation time step, and the total simulation time step covers the complete collection and transportation operation cycle. Within each simulation time step, the state variables of each collection and transportation vehicle are updated, including the current location, current load, current SOC, and list of completed garbage collection points. The vehicle's travel time is calculated by accumulating the predicted travel time of the collection and transportation vehicles.

[0031] After the simulation, the indicator calculation unit calculates multi-dimensional evaluation indicators for each candidate scheduling scheme based on the fusion model, including total collection and transportation time, total energy consumption, total charging waiting time, and total transfer station queuing time.

[0032] The evaluation report generation unit generates a complete simulation evaluation report for each candidate scheduling scheme. The report includes detailed routes for each vehicle, energy consumption curves for each time period, arrival time distribution for each transfer station, queuing heat map for charging stations, and comprehensive cost breakdown.

[0033] Furthermore, the collaborative scheduling module includes an objective function construction unit, a constraint modeling unit, an optimization algorithm solving unit, and a scheduling strategy output unit;

[0034] The objective function construction unit constructs a multi-objective optimization function based on the multi-dimensional evaluation indicators in the simulation evaluation report to minimize the weighted comprehensive cost;

[0035] The constraint modeling unit constructs constraints for candidate scheduling schemes;

[0036] The optimization algorithm solution unit uses an improved genetic algorithm to iteratively optimize all candidate scheduling schemes. The encoding method adopts a breakpoint-based permutation encoding, which represents the access order of each garbage point as a permutation. By inserting breakpoints, the permutation is divided into multiple sub-paths, and each sub-path is assigned to a car. After each generation of evolution, a local search is performed on the best individual in the current population: try to perform a 2-opt swap on the order of two garbage points in the path. If the overall cost of the new path is lower, the swap is accepted, and the current optimal scheduling scheme is output.

[0037] The scheduling strategy output unit decomposes the optimal scheduling scheme into a set of executable instructions, which are then distributed to the on-board terminals of each collection vehicle and the scheduling control console of the transfer station through the vehicle network platform.

[0038] The beneficial effects achieved by the present invention using the above solution are as follows:

[0039] (1) To address the problem of heterogeneity between path planning and dynamic traffic flow in traditional scheduling schemes, this scheme achieves high-precision time-varying traffic flow prediction through the iTransformer model and directly embeds the traffic flow speed matrix into the time-varying path modeling unit. It establishes an accurate and quantitative dynamic mapping between scheduling instructions and future road conditions. The improved Dijkstra algorithm is used to search for the shortest time path in the time-varying road network. The edge weights change dynamically with the time when vehicles enter the road segment, effectively avoiding predictive congestion. The BPR function is used to predict the travel time of the road segment, so that the path planning decision can respond to changes in traffic conditions in a forward-looking manner, rather than passively adapting to the congestion that has already occurred.

[0040] (2) In view of the problem that traditional scheduling schemes have experience differences when there are many garbage points and many vehicles, resulting in unstable scheduling quality, lack of simulation verification and multi-dimensional constraints, and difficulty in finding a globally optimal solution, this scheme realizes the automatic generation, simulation verification, multi-objective optimization and standardized execution of scheduling schemes. It uses the K-means clustering algorithm to generate a systematic initial solution for garbage points, solves the traveling salesman problem in the task subset of each vehicle, automatically generates a reasonable access order, and performs discrete-time simulation on each candidate scheduling scheme. Through dynamic optimization, it ensures the feasibility of the final output optimal scheme in actual execution, and significantly improves the intelligence level and operating efficiency of the urban garbage collection and transportation system. Attached Figure Description

[0041] Figure 1 This is a schematic diagram of the intelligent scheduling system for garbage transfer stations based on traffic flow prediction proposed in this invention.

[0042] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0043] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0044] Example 1, see Figure 1 The intelligent scheduling system for garbage transfer stations based on traffic flow prediction provided by this invention includes a data sensing module, a traffic flow prediction module, a fusion modeling module, a simulation analysis module, and a collaborative scheduling module.

[0045] The data perception module collects traffic flow data, operation and energy consumption data of collection vehicles, and operation data of garbage transfer stations and garbage collection points in residential areas within the collection responsibility area. The above data are sent to the traffic flow prediction module, the fusion modeling module and the simulation analysis module respectively.

[0046] The traffic flow prediction module builds a traffic flow prediction model based on the iTransformer architecture, uses historical traffic flow data to predict the traffic flow speed of each road segment on the road network in the future time period, outputs a traffic flow speed matrix and sends it to the fusion modeling module.

[0047] The fusion modeling module constructs a time-varying path model, a vehicle instantaneous energy consumption model, a time-varying queuing model, and a transfer service model based on the traffic flow velocity matrix, and couples them into a fusion model which is then sent to the simulation analysis module.

[0048] The simulation analysis module constructs a set of candidate scheduling schemes, uses a fusion model to simulate and extrapolate different candidate scheduling schemes, predicts multi-dimensional evaluation indicators, and generates a complete simulation evaluation report.

[0049] The collaborative scheduling module uses an improved genetic algorithm based on the simulation evaluation report to solve the optimal scheduling scheme for collection and transportation vehicles, and then distributes the solution to the on-board terminals of each collection and transportation vehicle and the scheduling control console of the transfer station.

[0050] Example 2, see Figure 1 This embodiment is based on the above embodiment, and the traffic flow prediction module includes a data preprocessing unit, a spatiotemporal feature extraction unit, and a model prediction unit;

[0051] The data preprocessing unit uses linear interpolation to fill in missing values ​​in historical traffic flow data, removes outliers based on the 3σ principle, and performs normalization processing. A fixed time step of 5 minutes is set, and the normalized data is aggregated into a time series matrix.

[0052] The spatiotemporal feature extraction unit models the road network as a graph, using each road segment as a node and the spatial connections between road segments as edges to construct a graph structure. Based on the time series matrix, it calculates the weights between adjacent nodes through a graph attention mechanism, aggregates the traffic state information of adjacent road segments to the current node, and extracts spatiotemporal features that fuse local spatial correlations. The formula used is as follows: ; ;

[0053] In the formula, Represents a road network map. This represents a set of nodes that includes all road segments in the road network. This represents an edge set that contains the spatial connections between all road segments. and Indicates the index of the node. Represents a node The spatiotemporal characteristics, Represents a node The set of neighboring nodes, Represents a node For nodes Attention coefficient This represents the learnable weight matrix. Represents a node The time series matrix, Indicates the activation function;

[0054] The model prediction unit constructs a traffic flow prediction model based on the iTransformer architecture. It inputs the spatiotemporal features of each road segment into the Transformer encoder, calculates the spatial correlation between different road segments through a self-attention mechanism, and predicts the traffic flow speed of each road segment in the future time period. The traffic flow prediction model is trained using the mean squared error loss function, with the Adam optimizer, an initial learning rate of 0.001, a batch size of 32, and 200 training epochs. The model is then aggregated into a traffic flow speed matrix output, using the following formula: ;

[0055] In the formula, This represents a matrix representing the predicted traffic flow speeds over a future time period. Represents a set of spatiotemporal feature sequences within a historical time window. This represents the nonlinear mapping function of the iTransformer model.

[0056] Example 3, see Figure 1 This embodiment is based on the above embodiment, and the fusion modeling module includes a travel time calculation unit, a time-varying path modeling unit, a vehicle energy consumption modeling unit, a charging queue modeling unit, a transfer service modeling unit, and a model fusion unit;

[0057] The travel time calculation unit fits the predicted traffic flow of each road segment to the traffic flow velocity matrix, and uses the BPR road resistance function to calculate the predicted travel time of collection vehicles through each road segment in each time period. The formula used is as follows: ;

[0058] In the formula, Indicates the road segment index. This indicates the time period index for when collection vehicles enter this road segment. Indicates the time period for collection vehicles Passing section Predicted travel time Indicates road segment Travel time of collection vehicles when there is no congestion. Indicates time period Time Section Predicted traffic flow Indicates road segment Road capacity and The parameters to be determined for the BPR road resistance function are 0.15 and 4.0, respectively.

[0059] The time-varying path modeling unit constructs a time-varying road network model based on the time when the collection vehicles enter the road segment and the predicted travel time in each time period. It uses an improved Dijkstra algorithm to search for the shortest time path from the starting point to the ending point in the time-varying road network model, calculates the earliest arrival time of each node, expands the nodes in the order of the earliest arrival times, and outputs the optimal path and its total travel time.

[0060] The vehicle energy consumption modeling unit establishes an instantaneous energy consumption model for the collection and transportation vehicles based on their load status and driving speed. The formula used is as follows: ;

[0061] In the formula, Indicates that the vehicle is on the road section Upper session Energy consumption during driving This indicates the base energy consumption rate when the vehicle is unloaded. Indicates the load-bearing energy consumption coefficient. Indicates the current load of the vehicle. The energy consumption coefficient represents the square of the velocity. Indicates road segment During the period Predicted traffic speed;

[0062] The charging queuing modeling unit establishes a time-varying queuing model based on the service capacity of charging stations. It fits the daily driving distance and travel probability distribution of the vehicles based on historical data on vehicle operation and energy consumption, and calculates the arrival rate of vehicles at charging stations using the following formula: ; ; ;

[0063] In the formula, Indicates a charging station. Indicates charging station During the period The arrival rate of collection and transportation vehicles This indicates the penetration rate of collection and transportation vehicles. Indicates the time from the start of the trip to the specified moment. Drive to charging station The cumulative driving distance, Indicates the maximum daily driving distance. Let be the probability density function of the daily travel distance. This represents the average queue length. This represents the average waiting time in the queue. Indicates charging station The number of charging stations Indicates the service rate per pile. Indicates the vehicle's battery capacity. This represents the target state of charge, set to 0.8. Indicates the state of charge upon arrival. Indicates charging power;

[0064] The transfer service modeling unit calculates the average waiting time at the waste transfer station for each time period and establishes a transfer service model. The formula used is as follows: ; ;

[0065] In the formula, Indicates a transit station. This represents the average waiting time. Indicates transit station During the period The arrival rate of collection and transportation vehicles This indicates the service rate per unloading station. This indicates the number of unloading positions at the transfer station. This indicates the busy rate of the transfer station service. Indicates the summation index;

[0066] The model fusion unit couples the time-varying path model, vehicle instantaneous energy consumption model, time-varying queuing model, and transfer service model into a fusion model.

[0067] By performing the aforementioned operations, this solution addresses the heterogeneity between path planning and dynamic traffic flow in traditional scheduling schemes. It achieves high-precision time-varying traffic flow prediction using the iTransformer model and directly embeds the traffic flow speed matrix into the time-varying path modeling unit. This establishes a precise and quantifiable dynamic mapping between scheduling instructions and future road conditions. An improved Dijkstra algorithm is employed to search for the shortest time path in the time-varying road network, with edge weights dynamically changing according to the time vehicles enter the road segment, effectively avoiding predictive congestion. The BPR function is used to predict the travel time of road segments, enabling path planning decisions to proactively respond to changes in traffic conditions rather than passively adapting to existing congestion.

[0068] Example 4, see Figure 1 This embodiment is based on the above embodiment, and the simulation analysis module includes a scheduling scheme generation unit, a simulation deduction unit, an index calculation unit, and an evaluation report generation unit;

[0069] The scheduling scheme generation unit uses the K-means clustering algorithm to divide the garbage points into groups based on the operation data, time window constraints, and number of available vehicles of each residential garbage point within the collection and transportation responsibility area. Each group corresponds to the task range of a collection and transportation vehicle. The traveling salesman problem is solved in the task subset of each collection and transportation vehicle to generate a preliminary set of candidate scheduling schemes. Each candidate scheduling scheme is a structured instruction set that specifies the garbage point access sequence, charging strategy, and transfer station arrival time for each collection and transportation vehicle.

[0070] The simulation and deduction unit performs discrete-time simulations for each candidate scheduling scheme, setting the simulation time step to 5 minutes. The total simulation time step covers the complete collection and transportation operation cycle. Within each simulation time step, the state variables of each collection and transportation vehicle are updated, including the current location, current load, current SOC, and list of completed garbage collection points. The vehicle's travel time is calculated by accumulating the predicted travel time of the collection and transportation vehicles.

[0071] After the simulation, the indicator calculation unit calculates multi-dimensional evaluation indicators for each candidate scheduling scheme based on the fusion model, including total collection and transportation time, total energy consumption, total charging waiting time, and total transfer station queuing time.

[0072] The evaluation report generation unit generates a complete simulation evaluation report for each candidate scheduling scheme. The report includes detailed routes for each vehicle, energy consumption curves for each time period, arrival time distribution for each transfer station, queuing heat map for charging stations, and comprehensive cost breakdown.

[0073] Example 5, see Figure 1 This embodiment is based on the above embodiment, and the collaborative scheduling module includes an objective function construction unit, a constraint modeling unit, an optimization algorithm solving unit, and a scheduling strategy output unit;

[0074] The objective function construction unit constructs a multi-objective optimization function based on the multi-dimensional evaluation indicators in the simulation evaluation report. The optimization objective is to minimize the weighted comprehensive cost from three dimensions: economic, time, and environmental. The formula used is as follows: ; ;

[0075] In the formula, Indicates the candidate scheduling scheme, Represents the comprehensive cost function. Represents the total economic cost. Indicates the total collection and transportation time. Indicates total energy consumption. This represents the revenue gained by responding to the demand-side response signals of the power grid. , , , These are the normalized weighting coefficients for the corresponding terms;

[0076] The constraint modeling unit constructs constraints for candidate scheduling schemes, including:

[0077] Full coverage constraint for collection and transportation tasks: Each garbage collection point must be visited at least once;

[0078] Vehicle load limit: The cumulative load of the vehicle at any time shall not exceed the rated load capacity;

[0079] Range constraint: The vehicle's SOC must not fall below the safety threshold on any road segment;

[0080] Time window constraint: For garbage collection points with time limits, vehicles must arrive within the specified time window;

[0081] Transfer station service capacity constraints: The number of vehicles served simultaneously at the same transfer station shall not exceed the number of unloading positions;

[0082] The optimization algorithm solution unit uses an improved genetic algorithm to iteratively optimize all candidate scheduling schemes. The encoding method adopts a breakpoint-based permutation encoding. The algorithm parameters are set as follows: population size is 200, maximum number of generations is 500, crossover probability is 0.85, and mutation probability is 0.1. The selection operation uses a tournament selection strategy with a size of 3, the crossover operation uses sequential crossover, and the mutation operation uses insertion mutation. After each generation, a local search is performed on the best individual in the current population: attempting to sequentially swap two garbage points in the path using 2-opt; if the overall cost of the new path is lower, the swap is accepted, and the current optimal scheduling scheme is output. The fitness function of the genetic algorithm is as follows: ;

[0083] In the formula, Indicate candidate scheduling schemes The fitness function;

[0084] The scheduling strategy output unit decomposes the optimal scheduling scheme into an executable instruction set, which is then distributed to the on-board terminals of each collection vehicle and the scheduling control console of the transfer station via the vehicle network platform. This includes:

[0085] Route planning instruction set: Generates detailed navigation routes for each collection vehicle;

[0086] Charging dispatch instruction set: clearly defines the charging station location, estimated arrival time, charging duration, and target SOC for each vehicle;

[0087] Arrival reservation instruction set: Send the estimated arrival time of each vehicle to each transfer station to achieve staggered and balanced arrival times.

[0088] By performing the aforementioned operations, this solution addresses the problems of traditional scheduling schemes, which suffer from inconsistent scheduling quality due to experience discrepancies when there are many garbage collection points and vehicles, lack of simulation verification and multi-dimensional constraints, and difficulty in finding a globally optimal solution. This solution automates the generation, simulation verification, multi-objective optimization, and standardized execution of scheduling schemes. It uses the K-means clustering algorithm to generate systematic initial solutions for garbage collection point allocation, solves the traveling salesman problem in the task subset of each vehicle to automatically generate a reasonable access order, and performs discrete-time simulations on each candidate scheduling scheme. Dynamic optimization ensures the feasibility of the final optimal solution in actual execution, significantly improving the intelligence level and operational efficiency of the urban waste collection and transportation system.

[0089] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0090] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

[0091] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. Intelligent scheduling system for waste transfer station based on traffic flow prediction, characterized in that: It includes a data perception module, a traffic flow prediction module, a fusion modeling module, a simulation analysis module, and a collaborative scheduling module; The data perception module collects traffic flow data, operation and energy consumption data of collection vehicles, and operation data of garbage transfer stations and garbage collection points in residential areas within the collection responsibility area. The above data are sent to the traffic flow prediction module, the fusion modeling module and the simulation analysis module respectively. The traffic flow prediction module builds a traffic flow prediction model based on the iTransformer architecture, uses historical traffic flow data to predict the traffic flow speed of each road segment on the road network in the future time period, outputs a traffic flow speed matrix and sends it to the fusion modeling module. The fusion modeling module constructs a time-varying path model, a vehicle instantaneous energy consumption model, a time-varying queuing model, and a transfer service model based on the traffic flow velocity matrix, and couples them into a fusion model which is then sent to the simulation analysis module. The simulation analysis module constructs a set of candidate scheduling schemes, uses a fusion model to simulate and extrapolate different candidate scheduling schemes, predicts multi-dimensional evaluation indicators, and generates a complete simulation evaluation report. The collaborative scheduling module uses an improved genetic algorithm based on the simulation evaluation report to solve the optimal scheduling scheme for collection and transportation vehicles, and then distributes the solution to the on-board terminals of each collection and transportation vehicle and the scheduling control console of the transfer station.

2. The intelligent scheduling system for garbage transfer station based on traffic flow prediction according to claim 1, characterized in that: The traffic flow prediction module includes a data preprocessing unit, a spatiotemporal feature extraction unit, and a model prediction unit. The data preprocessing unit performs missing value imputation, outlier removal and normalization on the original traffic flow data, and sets a fixed time step to aggregate the normalized data into a time series matrix; The spatiotemporal feature extraction unit models the road network as a graph, with each road segment in the road network as a node and the spatial connection relationship between road segments as edges, constructs a graph structure, calculates the weight between adjacent nodes based on the time series matrix through the graph attention mechanism, aggregates the traffic state information of adjacent road segments to the current node, and extracts spatiotemporal features that integrate local spatial correlation. The model prediction unit constructs a traffic flow prediction model based on the iTransformer architecture. It inputs the spatiotemporal features of each road segment into the Transformer encoder, calculates the spatial correlation between different road segments through a self-attention mechanism, predicts the traffic flow speed of each road segment in the future time period, and aggregates them into a traffic flow speed matrix for output. 3.The intelligent scheduling system for garbage transfer station based on traffic flow prediction of claim 2, characterized in that: The spatiotemporal feature extraction unit uses the following formula for spatiotemporal feature extraction: ； In the formula, and Indicates the index of the node. Represents a node The spatiotemporal characteristics, Represents a node The set of neighboring nodes, Represents a node For nodes Attention coefficient This represents the learnable weight matrix. Represents a node The time series matrix, This represents the activation function.

4. The intelligent scheduling system for garbage transfer station based on traffic flow prediction of claim 1, wherein: The fusion modeling module includes a travel time calculation unit, a time-varying path modeling unit, a vehicle energy consumption modeling unit, a charging queue modeling unit, a transfer service modeling unit, and a model fusion unit. The travel time calculation unit fits the predicted traffic flow of each road segment according to the traffic flow velocity matrix and uses the BPR road resistance function to calculate the predicted travel time of the collection vehicles through each road segment in each time period. The time-varying path modeling unit constructs a time-varying road network model based on the time when the collection vehicles enter the road segment and the predicted travel time in each time period. It uses an improved Dijkstra algorithm to search for the shortest time path from the starting point to the ending point in the time-varying road network model, calculates the earliest arrival time of each node, expands the nodes in the order of the earliest arrival times, and outputs the optimal path and its total travel time. The vehicle energy consumption modeling unit establishes an instantaneous energy consumption model of the vehicle based on the load status and driving speed of the collection and transportation vehicle. The charging queuing modeling unit establishes a time-varying queuing model based on the service capacity of the charging station. According to the historical operation and energy consumption data of the collection and transportation vehicles, it fits the daily driving distance and driving probability distribution of the collection and transportation vehicles, and calculates the arrival rate of the vehicles to the charging station. The transfer service modeling unit calculates the average waiting time at the waste transfer station for each time period and establishes a transfer service model; The model fusion unit couples the time-varying path model, vehicle instantaneous energy consumption model, time-varying queuing model, and transfer service model into a fusion model.

5. The intelligent scheduling system for garbage transfer station based on traffic flow prediction according to claim 4, characterized in that: In the travel time calculation unit, the formula for calculating the predicted travel time of collection vehicles through each road segment in each time period is as follows: ； In the formula, Indicates the road segment index. This indicates the time period index for when collection vehicles enter this road segment. Indicates the time period for collection vehicles Passing section Predicted travel time Indicates road segment Travel time of collection vehicles when there is no congestion. Indicates time period Time Section Predicted traffic flow Indicates road segment Road capacity and These are the undetermined parameters for the BPR path resistance function. 6.The intelligent scheduling system for garbage transfer station based on traffic flow prediction of claim 4, characterized in that: The simulation analysis module includes a scheduling scheme generation unit, a simulation deduction unit, an index calculation unit, and an evaluation report generation unit. The scheduling scheme generation unit uses the K-means clustering algorithm to divide the garbage points into groups based on the operation data, time window constraints, and number of available vehicles of each residential garbage point within the collection and transportation responsibility area. Each group corresponds to the task range of a collection and transportation vehicle. The traveling salesman problem is solved in the task subset of each collection and transportation vehicle to generate a preliminary set of candidate scheduling schemes. Each candidate scheduling scheme is a structured instruction set that specifies the garbage point access sequence, charging strategy, and transfer station arrival time for each collection and transportation vehicle. The simulation and deduction unit performs discrete-time simulation for each candidate scheduling scheme, sets the simulation time step, and the total simulation time step covers the complete collection and transportation operation cycle. Within each simulation time step, the state variables of each collection and transportation vehicle are updated, and the vehicle's travel time is calculated by accumulating the predicted travel time of the collection and transportation vehicles. After the simulation, the indicator calculation unit calculates multi-dimensional evaluation indicators for each candidate scheduling scheme based on the fusion model, including total collection and transportation time, total energy consumption, total charging waiting time, and total transfer station queuing time. The evaluation report generation unit generates a complete simulation evaluation report for each candidate scheduling scheme.

7. The intelligent scheduling system for garbage transfer station based on traffic flow prediction according to claim 1, characterized in that: The collaborative scheduling module includes an objective function construction unit, a constraint modeling unit, an optimization algorithm solving unit, and a scheduling strategy output unit; The objective function construction unit constructs a multi-objective optimization function based on the multi-dimensional evaluation indicators in the simulation evaluation report to minimize the weighted comprehensive cost; The constraint modeling unit constructs constraints for candidate scheduling schemes; The optimization algorithm solution unit uses an improved genetic algorithm to iteratively optimize all candidate scheduling schemes. The encoding method adopts a breakpoint-based permutation encoding, which represents the access order of each garbage point as a permutation. By inserting breakpoints, the permutation is divided into multiple sub-paths, and each sub-path is assigned to a car. After each generation of evolution, a local search is performed on the best individual in the current population: try to perform a 2-opt swap on the order of two garbage points in the path. If the overall cost of the new path is lower, the swap is accepted, and the current optimal scheduling scheme is output. The scheduling strategy output unit decomposes the optimal scheduling scheme into a set of executable instructions, which are then distributed to the on-board terminals of each collection vehicle and the scheduling control console of the transfer station through the vehicle network platform.

Images