A dynamic charging decision-making system and method linking shared mobility and charging services

Abstract

This invention relates to the field of shared mobility scheduling and charging service management technology, and particularly to a dynamic charging decision-making system and method that links shared mobility and charging services. The system includes: collecting multi-source data on vehicle battery level, charging pile status, and regional demand; analyzing historical data using a particle swarm optimization algorithm to generate scenario-based emergency charging thresholds and preventative charging thresholds; and performing noise reduction and standardization processing on real-time data. This invention effectively breaks the rigid limitation of traditional fixed battery thresholds on charging timing by constructing a deep matching mechanism that links offline parameter optimization with real-time data. It achieves precise linkage between the battery status of shared vehicles, travel demand, and charging resources, as well as a global optimization closed loop for the scheduling system, solving the dual balance problem of ensuring battery availability and fulfilling orders during operation.

Description

Technical Field

[0001] This invention relates to the field of shared mobility scheduling and charging service management technology, and in particular to a dynamic charging decision-making system and method that links shared mobility and charging services. Background Technology

[0002] With the rapid development of the sharing economy, shared electric bicycles and electric vehicles have become widely popular in urban core areas, peri-urban areas, and rural regions, serving as an important means to promote energy conservation and emission reduction and alleviate traffic congestion. In actual operation, battery management is a core factor restricting industry efficiency, as the adequacy of battery power directly determines the order service capacity and availability of transportation vehicles. To achieve overall optimization, operators need to monitor the vehicle's charge status in real time and deeply integrate it with fluctuating travel demand, dynamic charging resources, and scheduling strategies to ensure a precise balance between battery power assurance and order fulfillment rate across multiple scenarios.

[0003] However, existing charging decision-making technologies generally suffer from rigid strategies, relying primarily on fixed power thresholds to trigger charging. This leads to a severe disconnect between the decision-making process and real-time travel demand and charging station queue status. This disconnect makes charging vehicles unable to respond to sudden order demands, and because it does not consider power adaptability and real-time road conditions, it easily results in low utilization of charging resources or vehicles losing power midway. Summary of the Invention

[0004] To overcome the above shortcomings, this invention provides a dynamic charging decision-making system and method that links shared mobility with charging services, aiming to improve the problems of rigidity, disconnect between links, and low resource utilization in existing energy replenishment decision-making strategies.

[0005] In a first aspect, the present invention provides the following technical solution: a dynamic charging decision-making method linking shared mobility and charging services, comprising: S1. Collect multi-source data on vehicle battery level, charging pile status, and regional demand. Use particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventive charging thresholds. Perform noise reduction and standardization processing on real-time data. S2. Input real-time environmental features into the decision tree model to identify the current operating scenario, thereby automatically matching the corresponding emergency charging threshold and preventive charging threshold as a dynamic judgment benchmark for energy replenishment triggering. S3. By comparing the vehicle's battery level with the threshold and combining the demand intensity data predicted by the LSTM model for the next hour, the energy replenishment demand is determined as one of the highest priority, medium priority, or low priority. S4. Select and match power-adaptive charging piles within a 3km range based on priority needs, plan congestion-avoiding charging routes based on real-time road conditions and non-motorized vehicle lane traffic status, and push charging instructions containing location and priority to the terminal. S5. Based on monitoring feedback that orders surge by more than 50% and the satisfaction rate is less than 80%, execute the suspension of non-emergency tasks that have not started replenishing power or whose power level has not reached 50%, and execute the task resumption based on the monitoring signal that the satisfaction rate has rebounded to more than 90% and the intensity has dropped to below 1.0. S6 synchronizes charging progress and abnormal alarm information to the dispatch system in real time. Once the vehicle's battery level reaches the scenario-based target battery level threshold, charging is automatically terminated and the vehicle status is updated to available, completing the closed loop of linkage between charging services and travel needs.

[0006] Preferably, in step S1, the step of using particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventative charging thresholds includes: Initialize the particle swarm size and search space dimension, and encode each set of candidate emergency charging thresholds and preventive charging thresholds into the position vector of a single particle. Construct a joint fitness function with the objectives of maximizing order fulfillment rate, minimizing charging cost, and minimizing vehicle power outage rate, and calculate the individual extreme value and global extreme value of each particle at the current iteration number; The particle's velocity and spatial position are updated based on the individual and global extreme values. Through multiple iterations of optimization until the fitness function converges, the optimal threshold combination for urban core areas, urban-rural fringe areas, and rural areas is output.

[0007] Preferably, in step S1, the step of performing denoising and normalization processing on the real-time data includes: The data acquisition unit obtains real-time remaining battery power, latitude and longitude coordinates, real-time electricity price of charging piles, and historical order traffic data for each region, and uses the Laida criterion to remove outliers from the multi-source data. The cleaned multi-source data is mapped to the [0,1] interval using the min-max normalization method to generate a normalized real-time feature matrix; The normalized real-time feature matrix is ​​input into a relational database for index construction, and the output is a standardized real-time dataset to be called.

[0008] Preferably, in step S2, the step of inputting real-time environmental features into the decision tree model to identify the current operating scenario includes: Extract population density features, historical order peak time features, and geographic region attribute features from the standardized real-time dataset to construct a scene classification feature vector; The scene classification feature vector is input into a preset decision tree classifier, and the scene label to which the current geographic coordinates belong is determined by calculating the feature split layer by layer. Based on the scene tags, the corresponding emergency charging threshold and preventive charging threshold are retrieved and loaded from the pre-stored offline parameter table to complete the real-time dynamic matching of scene-based thresholds.

[0009] Preferably, in step S3, the step of determining the energy replenishment demand as one of the highest priority, medium priority, or low priority includes: The vehicle's real-time remaining battery power is compared with the emergency charging threshold. Based on the comparison results, vehicles with battery power below the threshold are directly classified as the highest priority tasks. Real-time regional traffic is input into the LSTM network, and the current demand intensity is determined by calculating the deviation rate between the order forecast value and the mean value for the next hour. Based on the determination that the battery level is between the two thresholds and the demand intensity exceeds the preset value, the vehicle is identified as a medium-priority task or a low-priority task.

[0010] Preferably, step S4, which involves selecting and matching power-adaptive charging pile locations within a 3km range, and planning congestion-avoidance energy replenishment routes based on real-time road conditions and non-motorized vehicle lane traffic status, includes: A search area with a radius of 3km is defined centered on the vehicle's current coordinates. A geofencing retrieval algorithm is used to obtain the real-time status of online charging stations within the area and to perform power compatibility filtering. The target matching stake location was determined using a multi-factor weighted scoring model based on distance, real-time electricity price, and number of people in the queue. The system retrieves the traffic status of non-motorized vehicle lanes and road network weight data for the target area, uses a path planning algorithm to search for the optimal congestion avoidance route, and pushes instructions including location and priority to the terminal.

[0011] Preferably, in step S5, the step of suspending non-emergency tasks that have not yet started replenishing power or whose power level has not reached 50% includes: The system iterates through the list of vehicles in the preventive charging or suspended charging levels and issues an immediate withdrawal order to vehicles in the suspended charging level that have not yet started charging. A forced power-off command is issued to vehicles that are in the preventive charging level and whose real-time battery level is less than 50%, and the vehicles are released into the order matching candidate pool after the command is executed. Update the status attributes of the vehicles whose missions were suspended to "idle and available" and synchronize their location information to the dispatch system to respond to sudden order demands.

[0012] Preferably, in step S5, the step of resuming the execution task includes: Continuously monitor regional order fulfillment rate and demand intensity data, and generate a task recovery signal when the fulfillment rate rises above 90% and the demand intensity drops below 1.0; The task recovery signal is sent to the dynamic adjustment unit, which is then instructed to retrieve the original priority sequence of the vehicles whose tasks were aborted. Re-execute the energy replenishment matching and planning according to the original priority order to restore the process of issuing and executing charging instructions to the suspended vehicles.

[0013] Preferably, in step S6, the step of synchronizing the charging progress and abnormal alarm information to the dispatch system, and automatically terminating charging after the vehicle's battery level reaches the scenario-based target battery level threshold, includes: The system monitors the vehicle's driving trajectory in real time. If the battery level drops below 50% of the emergency threshold while en route to a charging station, an abnormal alarm is sent and the nearest emergency charging point is planned. It receives pulse signals from charging piles and determines whether the vehicle's battery level has reached the scenario-specific target threshold of 80% in urban core areas or 90% in rural areas. After the termination conditions are met, a power supply stop message is sent and the cable lock is released, and the vehicle flag is updated to be available to complete the closed loop.

[0014] Secondly, the present invention provides the following technical solution: a dynamic charging decision-making system linking shared mobility and charging services, comprising: The data acquisition unit is used to collect multi-source data such as vehicle battery level, charging pile status and regional demand. It uses particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventive charging thresholds, and performs noise reduction and standardization processing on real-time data. The charging threshold determination unit is used to input real-time environmental characteristics into the decision tree model to identify the current operating scenario, thereby automatically matching the corresponding emergency charging threshold and preventive charging threshold as a dynamic determination benchmark for energy replenishment triggering. The charging priority ranking unit is used to determine the charging demand as one of the highest priority, medium priority, or low priority by comparing the vehicle's battery level with the threshold and combining the demand intensity data predicted by the LSTM model for the next hour. The charging strategy execution unit is used to filter and match power-adaptive charging piles within a 3km range based on priority needs, plan congestion-avoiding charging routes based on real-time road conditions and non-motorized vehicle lane traffic status, and push charging instructions containing location and priority to the terminal. The dynamic adjustment unit is used to suspend non-emergency tasks that have not started replenishing power or whose power level has not reached 50% based on monitoring feedback that the order surge exceeds 50% and the satisfaction rate is below 80%, and to resume tasks based on monitoring signals that the satisfaction rate has rebounded to above 90% and the intensity has dropped below 1.0. The status synchronization unit is used to synchronize the charging progress and abnormal alarm information to the dispatch system in real time. When the vehicle's battery level reaches the scenario-based target battery level threshold, it automatically terminates charging and updates the vehicle status to be available, thus completing the closed loop of linkage between charging services and travel needs.

[0015] The present invention has the following beneficial effects: 1. In this invention, by constructing a deep matching mechanism that combines offline parameter optimization with real-time data linkage, the rigid limitation of traditional fixed power threshold on the timing of energy replenishment is effectively broken. This enables precise linkage between the power status of shared vehicles, travel demand and charging resources, as well as a global optimization closed loop of the scheduling system, thus solving the dual balance problem of power supply guarantee and order fulfillment in operation.

[0016] 2. In this invention, a scenario-based three-level charging priority logic generated by particle swarm optimization algorithm is adopted. It can execute differentiated and precise energy replenishment strategies according to the real-time charge status of the vehicle, demand forecast and charging pile resource distribution. While reducing charging operation costs by more than 10%, it effectively avoids the waste of empty charging resources and significantly improves the economy and resource allocation efficiency of the energy replenishment system.

[0017] 3. In this invention, the dynamic adjustment unit flexibly adjusts the suspension and resumption of tasks during peak demand periods. The immediate cutoff logic for non-emergency power replenishment tasks ensures the maximum release of transportation resources in scenarios of order surges. While significantly improving the order fulfillment rate, the abnormal alarm and emergency power replenishment mechanism control the vehicle power outage rate at an extremely low level, thereby ensuring the continuity and stability of shared mobility services. Attached Figure Description

[0018] Figure 1 This is a flowchart of a dynamic charging decision-making method that links shared mobility and charging services, as proposed in this invention. Figure 2 This is an architecture diagram of a dynamic charging decision system that links shared mobility and charging services, as proposed in this invention. Figure 3 A flowchart illustrating the core process of dynamic charging decision-making in an urban core area. Detailed Implementation

[0019] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. Example 1

[0020] In a first embodiment of the present invention, the present invention provides a dynamic charging decision-making system and method that links shared mobility with charging services, such as... Figure 1 As shown, it includes the following steps: S1. Collect multi-source data on vehicle battery level, charging pile status, and regional demand. Use particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventive charging thresholds. Perform noise reduction and standardization processing on real-time data. Further, in step S1, the step of using particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventive charging thresholds includes: initializing the particle swarm size and search space dimension; encoding each set of candidate emergency charging thresholds and preventive charging thresholds as the position vector of a single particle; constructing a joint fitness function with the objectives of maximizing order fulfillment rate, minimizing charging cost, and minimizing vehicle power outage rate; calculating the individual extreme value and global extreme value of each particle at the current iteration number; updating the particle's motion speed and spatial position based on the individual extreme value and global extreme value; and iterating until the fitness function converges through multiple iterations to output the optimal threshold combination for urban core areas, urban-rural fringe areas, and rural areas.

[0021] Further, in step S1, the step of performing noise reduction and standardization processing on the real-time data includes: acquiring the vehicle's real-time remaining battery power, latitude and longitude coordinates, real-time electricity price of charging piles, and historical order traffic data of each region through the data acquisition unit, and using the Laida criterion to remove outlier points in the multi-source data; using the min-max normalization method to map the cleaned multi-source data to the [0,1] interval to generate a normalized real-time feature matrix; inputting the normalized real-time feature matrix into the relational database for index construction, and outputting the standardized real-time dataset to be called.

[0022] Specifically, the data acquisition unit first obtains multi-source data of shared transportation vehicles in real time through the IoT interface of the shared mobility service backend. This includes the vehicle's current battery level, Gaode latitude and longitude coordinates, estimated current driving range, and historical order traffic data for each operating area over the past ninety calendar days. Simultaneously, it retrieves the real-time location, interface power type, number of available charging stations, and current dynamic electricity price information for each charging station through a third-party charging service protocol interface. To eliminate outliers caused by sensor acquisition errors or communication packet loss, the system uses the Laida criterion to remove outliers from the aforementioned multi-source data. In specific calculations, a specific feature dataset sequence is used... First, calculate its arithmetic mean. and standard deviation If data points Satisfaction Criterion Formula If the value is found to be an outlier, the point will be identified as an abnormal outlier and a deletion operation will be performed.

[0023] The cleaned data is linearly transformed using the min-max normalization method, scaling it to a dimensionless range of zero to one, thereby eliminating the order-of-magnitude differences between features of different dimensions. The normalization calculation formula is expressed as: ; in, For the processed standardized data, For raw input data such as vehicle battery level or order demand intensity, and These represent the maximum and minimum values ​​of this feature in the historical observation samples, respectively. The processed, standardized real-time dataset is then input into a relational database, where an index structure based on timestamps and region numbers is established to facilitate high-speed retrieval by subsequent decision-making units.

[0024] After acquiring historical operational characteristics, the offline parameter optimization module uses a particle swarm optimization algorithm to iteratively calculate scenario-specific thresholds. The particle swarm size is set during the algorithm initialization phase. The number of dimensions in the search space is between fifty and one hundred. The value is set to two, corresponding to the emergency charging threshold to be optimized. and preventive charging threshold The position vector of each particle is represented as... The velocity vector is represented as To evaluate the merits of different threshold combinations, the system constructs a joint fitness function that integrates core indicators of shared mobility. The calculation formula is as follows: ; In the above formula, This represents the order fulfillment rate obtained through simulation using historical data. Represents the unit's energy replenishment cost. This represents the vehicle's power outage rate due to low battery levels. , , These are the preset weight coefficients. The goal of the algorithm is to find the fitness function that... The position of the particle that has reached its maximum value.

[0025] In each iteration, the particle is based on its individual historical best position. and the global optimal position of the group Update its own motion status using the following formula: ; ; In the interpretation of symbols, This represents the current iteration number. For inertial weights, and As a learning factor, and These are random numbers distributed between zero and one. After multiple iterations until the fitness function tends to stabilize and converge, the system finally outputs the optimal combination of threshold parameters for different scenarios such as urban core areas, urban-rural fringe areas, and rural areas. For example, the emergency charging threshold for urban core areas is 20%, the preventive charging threshold is 35%, while the threshold is automatically increased for rural areas with sparse facilities to ensure operational safety.

[0026] This step, through deep fusion and adaptive optimization of multi-source data, eliminates noise interference in operational data and provides accurate dynamic power replenishment benchmarks for different geographical scenarios. The implementation effect is demonstrated by a scientifically set dual-threshold system that significantly reduces the risk of vehicle power outages while substantially increasing available vehicle capacity during peak hours, achieving a balance between operating costs and service quality.

[0027] S2. Input real-time environmental features into the decision tree model to identify the current operating scenario, thereby automatically matching the corresponding emergency charging threshold and preventive charging threshold as a dynamic judgment benchmark for energy replenishment triggering. Further, in step S2, the step of inputting real-time environmental features into the decision tree model to identify the current operating scenario includes: extracting population density features, historical order peak time features, and geographical area attribute features from the standardized real-time dataset to construct a scenario classification feature vector; inputting the scenario classification feature vector into a preset decision tree classifier, and determining the scenario label to which the current geographical coordinates belong through layer-by-layer feature splitting calculation; based on the scenario label, retrieving and loading the corresponding emergency charging threshold and preventive charging threshold from the pre-stored offline parameter table to complete the real-time dynamic matching of scenario-based thresholds.

[0028] Specifically, the scene recognition unit first extracts environmental dimension features associated with the current geographic coordinates of the target vehicle from a standardized real-time dataset. These features include population density. Defined as the total resident population within a one-square-kilometer grid where the vehicle is currently located, and the historical peak time characteristics of orders. Defined by the weight of the starting hour with the highest order volume in the grid area over the past ninety days, based on geographical region attribute characteristics. Numerical encoding is used to represent administrative planning types such as commercial areas, residential areas, or suburbs. These feature components are combined to form a scene classification feature vector that describes the operating environment. .

[0029] The system inputs this feature vector into a pre-defined decision tree classifier, which selects splitting node features by calculating information gain during the construction phase. (Information gain) The calculation is based on information entropy The formula is expressed as: ; ; In the above formula, This represents the sample set of the operating environment. The total number of preset scene categories, such as urban core area, urban-rural fringe area and rural area. For the first The proportion of similar scenario samples in the set. For selected classification features such as population density, Features In the The decision tree classifier determines the subset of branch samples under each possible value. Whether the components in the data meet the node splitting threshold, for example, to determine... Whether the population density is greater than 15,000 people per square kilometer, the feature vector will ultimately be mapped to the scene label corresponding to the leaf node. .

[0030] The system uses the identified scene labels Retrieve the corresponding charging decision parameters from the pre-stored offline parameter table. This step loads the emergency charging threshold matching the current scenario using a primary key lookup. With preventative charging threshold For example, when the scene label When identified as the core urban area, the matching output It is 20%. The initial energy threshold is 35%. If the tag identifies a rural area, a higher energy threshold is automatically applied to address the sparse charging infrastructure. Finally, the dynamic parameters, based on the completed scenario matching, are pushed to the charging determination unit as a real-time benchmark for subsequent charging triggering.

[0031] This step achieves dynamic adaptation of charging triggering conditions through precise classification of the operating environment. The effect is that it allows for the setting of thresholds based on regional order intensity and differences in the distribution of replenishment resources, effectively balancing vehicle capacity redundancy and replenishment frequency in different geographical scenarios.

[0032] S3. By comparing the vehicle's battery level with the threshold and combining the demand intensity data predicted by the LSTM model for the next hour, the energy replenishment demand is determined as one of the highest priority, medium priority, or low priority. Furthermore, in step S3, the step of determining the energy replenishment demand as one of the highest priority, medium priority, or low priority includes: comparing the vehicle's real-time remaining battery power with the emergency charging threshold, and directly classifying vehicles with battery power below the threshold as the highest priority task based on the comparison result; inputting real-time regional traffic into the LSTM network, and determining the current demand intensity by calculating the offset rate between the predicted order value and the mean value for the next hour; and determining the vehicle as a medium priority task or a low priority task based on the determination result that the battery power is between the two thresholds and the demand intensity exceeds the preset value.

[0033] Specifically, the priority sorting unit first obtains the current state of charge of the vehicle to be decided. And extract the emergency charging threshold matched by the scene recognition module. The system performs a power comparison operation, and when the judgment formula is met... In such cases, the vehicle's energy replenishment needs are immediately designated as the highest priority, i.e., the emergency charging level, to ensure that the vehicle has basic power safety under any operating environment.

[0034] For vehicles whose remaining battery power is not in an emergency situation, the system uses a pre-set long short-term memory neural network model to predict the demand intensity of their respective operating grid for the next hour, and finally outputs the predicted order volume for the next hour. .

[0035] The system further calculates the demand intensity offset rate for the next hour. As a key indicator for determining environmental load, its calculation formula is expressed as: ; In this formula, The order forecast values ​​for the next hour output by the aforementioned model. Generate an average order value for this region over the same historical hour. The system will then calculate the offset rate. Compared with the preset intensity surge threshold and preventative charging threshold Perform a logical combination judgment. If the vehicle's battery level... satisfy And offset rate If the vehicle is deemed to face a future power outage risk and the region is about to experience a peak in orders, its charging demand is classified as medium priority, i.e., preventative charging. If the deviation rate... or vehicle battery level If the energy replenishment demand is determined to be of low priority, that is, the charging level is temporarily suspended, and the demand is released to the order matching module through the scheduling system to perform passenger carrying tasks.

[0036] This step, by deeply coupling the vehicle's internal battery status with external time-series predicted demand, transforms the energy replenishment decision-making process from a passive response to a proactive prevention. The results show that it can accurately identify potential high-demand scenarios and pre-select vehicles needing energy replenishment, significantly improving order fulfillment efficiency during peak periods while ensuring operational continuity.

[0037] S4. Select and match power-adaptive charging piles within a 3km range based on priority needs, plan congestion-avoiding charging routes based on real-time road conditions and non-motorized vehicle lane traffic status, and push charging instructions containing location and priority to the terminal. Further, in step S4, the steps of filtering and matching power-adaptive charging pile locations within a 3km range, and planning congestion-avoidance charging routes based on real-time road conditions and non-motorized vehicle lane traffic status, include: defining a search area with a radius of 3km centered on the vehicle's current coordinates; using a geofencing retrieval algorithm to obtain the real-time status of online charging stations within the area and performing power adaptability filtering; using a multi-factor weighted scoring model based on distance, real-time electricity price, and queue length to determine the target matching charging pile location; retrieving non-motorized vehicle lane traffic status and road network weight data for the target area; using a path planning algorithm to search for the optimal congestion-avoidance route and pushing instructions including location and priority to the terminal.

[0038] Specifically, the resource matching execution unit first obtains the current latitude and longitude location information of the vehicle to be recharged, and establishes a geofence search area with a radius of three kilometers centered on this location. The system uses a spatial indexing algorithm to retrieve the real-time status data of all online charging stations within this geofence, and retrieves the vehicle's charging protocol and rated voltage parameters. Based on protocol compatibility, the system performs power compatibility filtering on candidate stations, automatically eliminating charging stations that are offline or have incompatible interface types. For example, for the highest priority emergency charging task, the algorithm locks the filtering criteria to DC charging stations with fast charging interfaces, while for medium and low priority tasks, it focuses on filtering nearby AC slow charging stations.

[0039] After obtaining the set of suitable sites, the system uses a multi-factor weighted scoring model to calculate the comprehensive score of each candidate site. Input parameters include the physical distance from the vehicle to the station. The current price per kilowatt-hour at this station And the estimated waiting time The comprehensive scoring formula is expressed as follows: ; In the algorithm formula, and These are the minimum and maximum distances between candidate sites within the search range, respectively. and This represents the extreme value of electricity prices. and This represents the extreme value of queuing time. Weighting coefficients. , , The system dynamically allocates resources based on priority. For the highest priority task, the system increases the distance weight. and time weight To ensure vehicles arrive as quickly as possible, the electricity price weighting will be appropriately increased for low- to medium-priority tasks. To reduce operating costs, the charging pile with the highest score in the calculation results was identified as the unique target matching pile location.

[0040] After determining the target charging pile location, the system retrieves real-time traffic flow data and non-motorized vehicle lane traffic status characteristics within the operating area to construct a dynamically weighted road network topology map. During path calculation, the system maps factors such as non-motorized vehicle lane pavement quality, traffic width, and the presence of temporary obstacles to road segment toll cost weights. The path planning algorithm, based on the Dijkstra algorithm, searches the topology map for the path with the minimum cumulative weight from the vehicle's current point to the charging pile node. The formula is as follows: ; In the interpretation of symbols, For road section The physical length, The difficulty coefficient for non-motorized vehicle lanes. This represents the road congestion delay factor. The generated navigation coordinate sequence is then encapsulated with the target charging station identifier, priority level, and estimated refueling time into a charging instruction package, which is pushed to the vehicle's linked operation terminal via a wireless communication link.

[0041] This step, through precise resource selection and environmentally aware path intervention, achieves an efficient match between charging demand and physical infrastructure. The results demonstrate that it effectively avoids the risk of low-battery vehicles going to ineffective charging stations or congested road sections, significantly shortens charging turnaround time, and improves the resource utilization efficiency of charging piles.

[0042] S5. Based on monitoring feedback that orders surge by more than 50% and the satisfaction rate is less than 80%, execute the suspension of non-emergency tasks that have not started replenishing power or whose power level has not reached 50%, and execute the task resumption based on the monitoring signal that the satisfaction rate has rebounded to more than 90% and the intensity has dropped to below 1.0. Furthermore, in step S5, the step of suspending non-emergency tasks that have not started charging or whose battery level is less than 50% includes: traversing the list of vehicles in the preventive charging or suspended charging levels, issuing an immediate withdrawal command to vehicles in the suspended charging level that have not yet started charging; issuing a forced power-off command to vehicles in the preventive charging level that are charging and whose real-time battery level is less than 50%, and releasing the vehicles into the order matching candidate pool after the command is executed; updating the status attribute of the suspended task vehicle to idle and available, and synchronizing its location information to the scheduling system to respond to sudden order demands.

[0043] Furthermore, in step S5, the steps for performing task recovery include: continuously monitoring regional order fulfillment rate and demand intensity data, determining when the fulfillment rate rises to above 90% and the demand intensity drops below 1.0 to generate a task recovery signal; sending the task recovery signal to the dynamic adjustment unit and instructing it to retrieve the original priority sequence of the vehicles whose tasks were suspended; and re-executing the energy replenishment matching and planning according to the original priority order to restore the charging instruction issuance and execution process for the suspended vehicles.

[0044] Specifically, the dynamic adjustment unit monitors the load status of the operating area in real time through a sliding window algorithm. The key input parameters include the number of orders generated in real time within the current statistical time window. and the average number of orders generated in the past hour The system uses formulas Calculate the demand surge rate and simultaneously obtain the real-time order fulfillment rate fed back by the scheduling module. When the monitoring feedback meets the judgment logic and When this occurs, the system immediately identifies it as a sudden peak demand scenario and generates a trigger signal, initiating the termination process for non-emergency power replenishment tasks.

[0045] Upon receiving the trigger signal, the system automatically iterates through the list of all non-highest priority tasks, i.e., the list of vehicles in the preventative charging or deferred charging levels. For vehicles in the deferred charging level that have not yet arrived at a charging station to begin the physical charging process, the system pushes an immediate withdrawal command to their linked operating terminal, directly clearing the navigation path and terminating the current charging assignment via the command packet. For vehicles already in the preventative charging level and in the charging state, the system reads their state of charge values ​​in real time through the charging protocol interface. If the inequality is satisfied... The system then sends a forced power-off command message to the charging pile, disconnecting the current supply by driving the physical layer relay; if If the vehicle is charged, the charging task will be retained to maintain the vehicle's basic range.

[0046] After the abort operation is completed, the system immediately updates the cloud attributes of the vehicles whose tasks were aborted through the status synchronization unit, resetting the vehicle status field from charging to idle and available. At the same time, the system forcibly triggers the vehicle positioning module to transmit the latest latitude and longitude coordinates and synchronizes its location data in real time to the order matching candidate pool of the hybrid intelligent dispatch system. This allows these vehicles to participate in the nearest order dispatch logic while in an idle state, thereby achieving instantaneous replenishment of regional transportation capacity.

[0047] The dynamic adjustment unit continuously monitors the recovery of environmental indicators during the optimization process, with input parameters including the adjusted real-time order fulfillment rate. and demand intensity coefficient When the system monitoring signal meets the judgment condition. and When the system automatically issues a task recovery signal, the dynamic adjustment unit retrieves the original priority sequence tags of the suspended vehicles from the cache stack based on this signal, maintaining their original urgency order. Subsequently, the system re-triggers the screening and planning algorithm in step four according to the original priority order, re-matching the optimal charging station for vehicles that meet the recovery conditions and pushing the energy replenishment navigation command, thus completing a smooth transition from emergency dispatch state to regular energy replenishment logic.

[0048] This step, through a flexible adjustment logic driven by multiple indicators, achieves a second-level rebalancing of energy replenishment resources and travel demand in scenarios with extremely limited capacity. The implementation effect is demonstrated by effectively mitigating peak order pressure and maximizing the overall order service efficiency of the region while ensuring a minimum core power supply.

[0049] S6 synchronizes charging progress and abnormal alarm information to the dispatch system in real time. Once the vehicle's battery level reaches the scenario-based target battery level threshold, charging is automatically terminated and the vehicle status is updated to available, completing the closed loop of linkage between charging services and travel needs.

[0050] Furthermore, in step S6, the steps of synchronizing the charging progress and abnormal alarm information to the dispatch system and automatically terminating charging after the vehicle's battery level reaches the scenario-based target battery level threshold include: real-time monitoring of the vehicle's driving trajectory; if the battery level is lower than 50% of the emergency threshold while on the way to charging, an abnormal alarm is pushed and the nearest emergency charging point is planned; receiving pulse signals from the charging pile to determine whether the vehicle's battery level has reached the scenario-based target threshold of 80% in urban core areas or 90% in rural areas; and after the termination conditions are met, issuing a stop power supply message and triggering the release of the cable lock, and updating the vehicle's flag to be available to complete the closed loop.

[0051] Specifically, the state synchronization unit first obtains the current state of charge of the recharging vehicle in real time through high-frequency heartbeat packets. The system uses its geographic coordinates to enable trajectory tracking and monitoring. To prevent power outages caused by vehicle exhaustion while en route to a charging station, the system continuously determines whether the vehicle's remaining battery power meets the abnormal alarm criteria. ; Symbol interpretation This displays the vehicle's current real-time remaining battery percentage. This represents the emergency charging threshold obtained from step two. When the calculation result... Upon that, the system immediately sends an emergency alarm message to the operation terminal and calls the path planning engine to re-retrieve the search radius. The nearest emergency power supply point within the area is located to ensure escort safety.

[0052] During the charging process, the system continuously receives periodically transmitted power pulse signals through the charging pile communication protocol interface and synchronizes them in real time to the scheduling system's database. The system then uses the scene tags identified in step two to further refine the process. Dynamically load scenario-based target power threshold The task exit process is triggered when the termination criteria are met. The termination logic follows a segmented function: ; An output of 1 indicates that the termination condition has been met; this is specifically set for scenarios identified as urban core areas. The percentage is 80%, while the setting is for scenarios identified as rural areas. It is 90%.

[0053] Once the aforementioned power termination conditions are met, the system sends a standardized power supply stop message command to the charging pile control module and sends a control signal to the electromagnetic lock mechanism to trigger the release action of the cable locking mechanism. Subsequently, the system updates the vehicle status field in the shared mobility scheduling system and order matching module from "charging" to "idle and available," and re-includes the vehicle in the global order dispatch candidate pool to complete the closed loop of linkage between charging service and travel demand. This step ensures the safety of the charging process and realizes the timely recovery of transportation capacity resources through precise end-point control and real-time status synchronization. Example 2

[0054] Existing charging decision-making technologies generally suffer from rigid strategies, relying primarily on fixed energy thresholds to trigger charging. This leads to a severe disconnect between the decision-making process and real-time travel demand and charging station queuing status. To address these issues, this invention provides a dynamic charging decision-making system that links shared mobility with charging services. Its structure is as follows: Figure 2 As shown, it includes: The data acquisition unit is used to collect multi-source data such as vehicle battery level, charging pile status and regional demand. It uses particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventive charging thresholds, and performs noise reduction and standardization processing on real-time data. The charging threshold determination unit is used to input real-time environmental characteristics into the decision tree model to identify the current operating scenario, thereby automatically matching the corresponding emergency charging threshold and preventive charging threshold as a dynamic determination benchmark for energy replenishment triggering. The charging priority ranking unit is used to determine the charging demand as one of the highest priority, medium priority, or low priority by comparing the vehicle's battery level with the threshold and combining the demand intensity data predicted by the LSTM model for the next hour. The charging strategy execution unit is used to filter and match power-adaptive charging piles within a 3km range based on priority needs, plan congestion-avoiding charging routes based on real-time road conditions and non-motorized vehicle lane traffic status, and push charging instructions containing location and priority to the terminal. The dynamic adjustment unit is used to suspend non-emergency tasks that have not started replenishing power or whose power level has not reached 50% based on monitoring feedback that the order surge exceeds 50% and the satisfaction rate is below 80%, and to resume tasks based on monitoring signals that the satisfaction rate has rebounded to above 90% and the intensity has dropped below 1.0. The status synchronization unit is used to synchronize the charging progress and abnormal alarm information to the dispatch system in real time. When the vehicle's battery level reaches the scenario-based target battery level threshold, it automatically terminates charging and updates the vehicle status to be available, thus completing the closed loop of linkage between charging services and travel needs.

[0055] Specifically, this system first acquires dynamic data such as real-time vehicle battery level, location coordinates, and charging pile occupancy through a data acquisition unit. It then uses the Raida criterion to identify and remove outliers exceeding three standard deviations. Next, it performs min-max normalization to generate a standardized feature matrix. Simultaneously, it uses a particle swarm optimization algorithm to iteratively optimize historical data from the past ninety days. By constructing a fitness function centered on order fulfillment rate, charging cost, and power outage rate, it calculates the optimal dual-threshold combination parameters for different geographical scenarios. and Subsequently, the charging threshold determination unit extracts the population density and peak features of the current grid, inputs them into the decision tree model for feature splitting calculation to identify the operational scenario, thereby automatically loading the corresponding dynamic judgment benchmark for the current decision area; in the priority ranking stage, the system will determine the vehicle's current state of charge. The highest priority tasks are identified by comparing them with an emergency threshold, and an LSTM model is used to predict the order offset rate for the next hour. Based on whether the offset rate exceeds a preset intensity value, vehicles within the dual thresholds are classified into medium-priority or low-priority tasks. The charging strategy execution unit executes interface protocol and power adaptability filtering within a three-kilometer range, applies a multi-factor weighted scoring model based on distance, electricity price, and queue length to select target charging piles, and plans congestion-avoiding charging routes by combining real-time road conditions and non-motorized vehicle lane toll costs. The dynamic adjustment unit monitors the regional load in real time. If the order volume surges by 50% and the satisfaction rate is below 80%, a withdrawal command is issued to suspend the charging task, and the charging vehicles with less than 50% battery power are forcibly disconnected to release capacity to the candidate pool. The task is resumed when the satisfaction rate rises to above 90% and the demand intensity drops below 1.0. The status synchronization unit performs trajectory tracking throughout the charging process. If the battery power drops to the alarm line en route, an emergency charging point is planned in real time. After the vehicle battery power reaches the scenario-based target threshold, a stop power supply command is automatically issued and the cable lock is released, and the vehicle status is updated to be available to complete the scheduling loop.

[0056] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A dynamic charging decision-making method linking shared mobility and charging services, characterized in that, Includes the following steps: S1. Collect multi-source data on vehicle battery level, charging pile status, and regional demand. Use particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventive charging thresholds. Perform noise reduction and standardization processing on real-time data. S2. Input real-time environmental features into the decision tree model to identify the current operating scenario, thereby automatically matching the corresponding emergency charging threshold and preventive charging threshold as a dynamic judgment benchmark for energy replenishment triggering. S3. By comparing the vehicle's battery level with the threshold and combining the demand intensity data predicted by the LSTM model for the next hour, the energy replenishment demand is determined as one of the highest priority, medium priority, or low priority. S4. Select and match power-adaptive charging piles within a 3km range based on priority needs, plan congestion-avoiding charging routes based on real-time road conditions and non-motorized vehicle lane traffic status, and push charging instructions containing location and priority to the terminal. S5. Based on monitoring feedback that orders surge by more than 50% and the satisfaction rate is less than 80%, execute the suspension of non-emergency tasks that have not started replenishing power or whose power level has not reached 50%, and execute the task resumption based on the monitoring signal that the satisfaction rate has rebounded to more than 90% and the intensity has dropped to below 1.

0. S6 synchronizes charging progress and abnormal alarm information to the dispatch system in real time. Once the vehicle's battery level reaches the scenario-based target battery level threshold, charging is automatically terminated and the vehicle status is updated to available, completing the closed loop of linkage between charging services and travel needs.

2. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, In step S1, the step of using particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventative charging thresholds includes: Initialize the particle swarm size and search space dimension, and encode each set of candidate emergency charging thresholds and preventive charging thresholds into the position vector of a single particle. Construct a joint fitness function with the objectives of maximizing order fulfillment rate, minimizing charging cost, and minimizing vehicle power outage rate, and calculate the individual extreme value and global extreme value of each particle at the current iteration number; The particle's velocity and spatial position are updated based on the individual and global extreme values. Through multiple iterations of optimization until the fitness function converges, the optimal threshold combination for urban core areas, urban-rural fringe areas, and rural areas is output.

3. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, In step S1, the step of performing denoising and normalization processing on the real-time data includes: The data acquisition unit obtains real-time remaining battery power, latitude and longitude coordinates, real-time electricity price of charging piles, and historical order traffic data for each region, and uses the Laida criterion to remove outliers from the multi-source data. The cleaned multi-source data is mapped to the [0,1] interval using the min-max normalization method to generate a normalized real-time feature matrix; The normalized real-time feature matrix is ​​input into a relational database for index construction, and the output is a standardized real-time dataset to be called.

4. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, Step S2, the step of inputting real-time environmental features into the decision tree model to identify the current operating scenario, includes: Extract population density features, historical order peak time features, and geographic region attribute features from the standardized real-time dataset to construct a scene classification feature vector; The scene classification feature vector is input into a preset decision tree classifier, and the scene label to which the current geographic coordinates belong is determined by calculating the feature split layer by layer. Based on the scene tags, the corresponding emergency charging threshold and preventive charging threshold are retrieved and loaded from the pre-stored offline parameter table to complete the real-time dynamic matching of scene-based thresholds.

5. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, Step S3, the step of determining the energy replenishment demand as one of the highest priority, medium priority, or low priority, includes: The vehicle's real-time remaining battery power is compared with the emergency charging threshold. Based on the comparison results, vehicles with battery power below the threshold are directly classified as the highest priority tasks. Real-time regional traffic is input into the LSTM network, and the current demand intensity is determined by calculating the deviation rate between the order forecast value and the mean value for the next hour. Based on the determination that the battery level is between the two thresholds and the demand intensity exceeds the preset value, the vehicle is identified as a medium-priority task or a low-priority task.

6. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, Step S4 involves selecting and matching power-adaptive charging pile locations within a 3km range, and planning congestion-avoidance energy replenishment routes based on real-time road conditions and non-motorized vehicle lane traffic status. This includes: A search area with a radius of 3km is defined centered on the vehicle's current coordinates. A geofencing retrieval algorithm is used to obtain the real-time status of online charging stations within the area and to perform power compatibility filtering. The target matching stake location was determined using a multi-factor weighted scoring model based on distance, real-time electricity price, and number of people in the queue. The system retrieves the traffic status of non-motorized vehicle lanes and road network weight data for the target area, uses a path planning algorithm to search for the optimal congestion avoidance route, and pushes instructions including location and priority to the terminal.

7. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, Step S5, the step of suspending non-emergency tasks that have not yet started charging or whose battery level has not reached 50%, includes: The system iterates through the list of vehicles in the preventive charging or suspended charging levels and issues an immediate withdrawal order to vehicles in the suspended charging level that have not yet started charging. A forced power-off command is issued to vehicles that are in the preventive charging level and whose real-time battery level is less than 50%, and the vehicles are released into the order matching candidate pool after the command is executed. Update the status attributes of the vehicles whose missions were suspended to "idle and available" and synchronize their location information to the dispatch system to respond to sudden order demands.

8. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, Step S5, the step of resuming the execution task, includes: Continuously monitor regional order fulfillment rate and demand intensity data, and generate a task recovery signal when the fulfillment rate rises above 90% and the demand intensity drops below 1.0; The task recovery signal is sent to the dynamic adjustment unit, which is then instructed to retrieve the original priority sequence of the vehicles whose tasks were aborted. Re-execute the energy replenishment matching and planning according to the original priority order to restore the process of issuing and executing charging instructions to the suspended vehicles.

9. The dynamic charging decision-making method for linking shared mobility and charging services according to claim 1, characterized in that, Step S6, which involves synchronizing the charging progress and abnormal alarm information to the dispatch system, and automatically terminating charging after the vehicle's battery level reaches the scenario-based target battery threshold, includes: The system monitors the vehicle's driving trajectory in real time. If the battery level drops below 50% of the emergency threshold while en route to a charging station, an abnormal alarm is sent and the nearest emergency charging point is planned. It receives pulse signals from charging piles and determines whether the vehicle's battery level has reached the scenario-specific target threshold of 80% in urban core areas or 90% in rural areas. After the termination conditions are met, a power supply stop message is sent and the cable lock is released, and the vehicle flag is updated to be available to complete the closed loop.

10. A dynamic charging decision system linking shared mobility and charging services, used in the dynamic charging decision method linking shared mobility and charging services as described in any one of claims 1-9, characterized in that, include: The data acquisition unit is used to collect multi-source data such as vehicle battery level, charging pile status and regional demand. It uses particle swarm optimization algorithm to analyze historical data to generate scenario-based emergency charging thresholds and preventive charging thresholds, and performs noise reduction and standardization processing on real-time data. The charging threshold determination unit is used to input real-time environmental characteristics into the decision tree model to identify the current operating scenario, thereby automatically matching the corresponding emergency charging threshold and preventive charging threshold as a dynamic determination benchmark for energy replenishment triggering. The charging priority ranking unit is used to determine the charging demand as one of the highest priority, medium priority, or low priority by comparing the vehicle's battery level with the threshold and combining the demand intensity data predicted by the LSTM model for the next hour. The charging strategy execution unit is used to filter and match power-adaptive charging piles within a 3km range based on priority needs, plan congestion-avoiding charging routes based on real-time road conditions and non-motorized vehicle lane traffic status, and push charging instructions containing location and priority to the terminal. The dynamic adjustment unit is used to suspend non-emergency tasks that have not started replenishing power or whose power level has not reached 50% based on monitoring feedback that the order surge exceeds 50% and the satisfaction rate is below 80%, and to resume tasks based on monitoring signals that the satisfaction rate has rebounded to above 90% and the intensity has dropped below 1.

0. The status synchronization unit is used to synchronize the charging progress and abnormal alarm information to the dispatch system in real time. When the vehicle's battery level reaches the scenario-based target battery level threshold, it automatically terminates charging and updates the vehicle status to be available, thus completing the closed loop of linkage between charging services and travel needs.

Images