Traffic-power fusion network bearing capacity analysis method based on double-layer joint simulation

Abstract

The invention discloses a traffic-electric power fusion network bearing capacity analysis method based on a double-layer model and multi-software joint simulation, and relates to the technical field of traffic engineering and electric power engineering crossing. According to the technical scheme, an upper-layer system optimization-lower-layer operation simulation double-layer model normal form is introduced, the upper layer aims at maximizing the total bearing capacity of the traffic-electric power fusion network, the lower layer constructs a real-time interaction channel through a TraCI interface of SUMO and a TCP / IP tool box of Matlab, and dynamic coupling simulation of the two networks is achieved; and solving the model by adopting a Bayesian optimization algorithm, and verifying the precision of the model through actual scene data in combination with a three-level evaluation index system covering traffic, electric power and fusion collaborative dimensions. The problems that a traditional method is insufficient in evaluation precision and poor in collaboration are solved, precise evaluation of the bearing capacity of the fusion network is achieved, technical support can be directly provided for charging facility layout and two-network collaborative scheduling in highways and other scenes, and the engineering application value is remarkable.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traffic engineering and power engineering, and particularly relates to a traffic network and power network fusion carrying capacity analysis method fusing a bilevel model paradigm, SUMO-Matlab joint simulation and simulation optimization solution, which is suitable for fusion network carrying capacity evaluation, planning and design and operation scheduling optimization containing electric vehicle charging scenes. BACKGROUND

[0002] Traditional analysis methods of traffic network carrying capacity take analytical models as cores, solve the maximum carrying demand of the road network through the construction of network capacity models, traffic flow distribution models and other mathematical optimization models. The core constraints of such methods focus on single network level indicators such as traffic flow conservation and road network passing efficiency, and do not involve cross-domain constraint association. Power network carrying capacity analysis also takes analytical methods as the main technical means, and carries out evaluation through tools such as power flow calculation analysis model and load distribution optimization model, and its core focuses on power system supply-demand balance state and power transmission channel capacity and other power network exclusive constraint boundaries.

[0003] With the large-scale popularization and application of electric vehicles, the charging demand of the traffic network and the power supply capacity of the power network form a closer and closer coupling association, and the traditional single network analytical model cannot accurately depict the cross-network dynamic interaction characteristics. In the existing technology, although some studies introduce SUMO (Simulation of Urban Mobility) traffic simulation tools and Matlab distribution network simulation tools to carry out related analysis, such studies mostly adopt the data splicing mode after independent simulation, and do not establish a real-time interaction mechanism between the two networks. At the same time, for the optimization problem of traffic-power fusion network carrying capacity, there is currently a lack of mature "upper system optimization-lower operation constraint" bilevel model paradigm support, and the traditional analytical solution method is difficult to adapt to the complex constraint conditions and dynamic coupling relationship in the simulation scene, resulting in insufficient precision of the fusion network carrying capacity evaluation results and limited engineering application value.

[0004] At present, the existing defects in the prior art are as follows: 1. The joint simulation collaboration is missing: SUMO and Matlab simulation are in an independent running state, and a real-time data interaction channel is not built based on a reliable communication protocol. The constraint of the power supply state of the distribution network on the charging behavior of the electric vehicle cannot be dynamically fed back, and the impact of the real-time change of the charging demand on the load of the distribution network cannot be reflected in real time. Therefore, there is a significant deviation between the analysis result and the actual fusion scenario. Taking the scenario of concentrated charging of electric vehicles in the peak period as an example, the existing technology cannot feed back the operation state of the overload of the distribution network line load rate to the traffic simulation system, nor can it transmit the real-time fluctuation of the charging demand of the electric vehicle to the power simulation system. Moreover, the existing technology does not clearly define the feasibility and specific implementation path of the TCP / IP protocol in the simulation interaction of the two networks, making it difficult to implement the collaborative simulation scheme in engineering.

[0005] 2. The model paradigm and the solving method are not suitable: The dual-layer model paradigm is not used to describe the core optimization logic of the fusion network carrying capacity, and the hierarchical relationship between the upper target of "system level carrying capacity maximization" and the lower demand of "satisfaction of the operation constraints of the two networks" cannot be clearly defined. At the same time, the traditional analytical solving method is difficult to handle the dynamic coupling constraints in the simulation scenario, such as the randomness of the charging behavior of the electric vehicle due to the difference in the remaining power and the charging price, and the load fluctuation of the distribution network due to the fluctuation of residential electricity and industrial electricity superimposed with the charging load, resulting in low efficiency and insufficient accuracy in the solving process.

[0006] 3. The limitations of traditional analytical models are highlighted: Early traffic network carrying capacity analytical models only take the road network traffic efficiency and traffic flow capacity as the core evaluation dimensions, without considering the power network constraints such as power supply capacity and voltage stability. The power network analytical model focuses on power transmission capacity, supply and demand balance, and other power system indicators, without considering the impact of electric vehicle charging demand on the distribution network. Taking the charging scenario at a highway service area as an example, the traditional traffic analytical model only calculates the upper limit of the traffic carrying capacity corresponding to the service capacity of the charging facility, without considering the feeder load rate constraint of the distribution network connected to the service area. The traditional power analytical model only calculates the carrying capacity of the distribution network according to the fixed load, without considering the dynamic charging load formed by the change of traffic flow, which cannot accurately simulate the dynamic interaction process between electric vehicle charging decisions, road congestion and distribution network load fluctuation, and cannot fully reflect the actual boundary of the fusion network carrying capacity. SUMMARY

[0007] To solve the problems of joint simulation collaboration, model paradigm and solving method not suitable, and limitations of traditional analytical models in the prior art, the present application clearly defines the real-time interaction feasibility of SUMO and Matlab based on the TCP / IP protocol, introduces a dual-layer model paradigm, and combines a simulation optimization solving method to build an integrated analysis method that fuses "model framework-joint simulation-high efficiency solving", and realizes accurate and comprehensive evaluation of the traffic-power fusion network carrying capacity.

[0008] Specifically as follows: A traffic-power integrated network carrying capacity analysis method based on double-layer joint simulation, comprising the following steps: Step S1: constructing a double-layer model framework including upper and lower layer models, and specifying the objectives and constraints of the upper and lower layer models; The upper layer model is a system optimization layer, and the objective is to maximize the total carrying capacity of the integrated network, which is specifically to maximize the matching value of the total OD demand of the traffic network and the safe power supply capacity of the power network; the constraint conditions include the threshold value of the average travel time of the traffic network, the upper limit of the carbon emission of the power network, and the upper limit of the service time length of the charging facility, and the constraint system design refers to the classical constraint framework of the double-layer model for traffic-power integrated network carrying capacity analysis; The lower layer model is a running simulation layer, and the objective is to satisfy the dynamic running constraints of the traffic network and the power network, and the joint simulation of SUMO and Matlab is realized; the constraint conditions include the traffic flow conservation constraint, the voltage deviation threshold value control of the distribution network within ±5%, the upper limit of the line load rate set to 85%, and the electric vehicle charging power constraint, which dynamically depicts the traffic-power coupled running state through the above constraints; Step S2: building a real-time interaction system of SUMO-Matlab based on TCP / IP protocol; Realize the engineering application of TCP / IP protocol in the simulation interaction of the two networks: build a two-way data interaction channel relying on the TraCI interface of SUMO and the TCP / IP toolbox integrated with Matlab, realize the dynamic feedback of charging demand and power supply constraint signals, wherein the TraCI interface originally supports the TCP / IP communication protocol, the TCP / IP toolbox completes the data receiving and sending configuration, and through the channel, the real-time and reliability of data transmission between the two networks are ensured, which provides communication support for subsequent dynamic coupling simulation; Formulate the data format and transmission rules: in the data sending end, SUMO collects and outputs two types of core data in real time, one type is electric vehicle charging request data, specifically including charging facility identification, charging demand power and expected charging time length, and the other type is road network running state data, specifically including traffic flow of each road section and average vehicle speed of the road network, and the above data is transmitted to Matlab after being packaged by TCP / IP protocol; in the data receiving and feedback end, Matlab receives the data and carries out real-time power flow calculation of the distribution network to obtain the running state parameters such as node voltage and line load rate of the distribution network, compares the calculation results with the preset constraint threshold value, generates power supply constraint signals if the threshold value is triggered, and feeds back to SUMO through the same interaction channel, and SUMO adjusts the running parameters of the charging facility or the driving path of the electric vehicle according to the signals; Step S3: Construct a lower-level SUMO-Matlab co-simulation model, in which a traffic simulation model containing the Logit charging decision algorithm is constructed in the SUMO traffic simulation model, and a distribution network simulation model containing charging load is constructed in the Matlab distribution network simulation model, so as to realize the dynamic coupling simulation of the two networks. SUMO Traffic Simulation Model: As the core traffic-side module of the lower-level co-simulation, it imports road network structure, OD matrix and electric vehicle parameters, and embeds a charging decision algorithm. This charging decision algorithm adopts the Logit model. During the model construction process, it comprehensively considers the distance between charging facilities and vehicles and the time preference of vehicles traveling to charging facilities. When receiving power supply constraint signals from Matlab, the model responds in real time and executes adjustment strategies, specifically including dynamically reducing the operating power of charging facilities or guiding vehicles waiting to be charged to plan their travel paths to other compliant charging facilities, so as to achieve dynamic matching between traffic operation status and power constraints. Matlab Distribution Network Simulation Model: As the core module of the power side in the lower-level co-simulation, it imports the distribution network topology, line parameters, and baseline load data; after receiving the electric vehicle charging demand data output by SUMO through the TCP / IP interactive channel, it uses it as a dynamic load to access the distribution network simulation scenario, and uses the Newton-Raphson method to perform real-time power flow calculations to obtain core operating parameters such as distribution network node voltage and line load rate. These parameters are then compared with preset power constraint thresholds to complete compliance judgments and provide data support for subsequent constraint signal feedback. Step S4: Solve the two-layer model using simulation optimization methods, construct the surrogate model through Gaussian process regression, and iteratively optimize based on the expected improvement criterion to obtain the optimal solution for the fusion bearing capacity; The solution strategy is determined by simulation optimization method, and the Bayesian optimization algorithm is selected as the solver of the two-layer model. The solver takes the objective function of maximizing the bearing capacity of the upper-layer model as the optimization direction, and takes the satisfaction of the two-network operation constraints output by the lower-layer SUMO-Matlab co-simulation as the feedback basis, forming a closed-loop solution logic of "upper-layer optimization - lower-layer simulation feedback". Step S5: Construct a comprehensive bearing capacity assessment index system, verify the model using actual data, and classify bearing capacity levels; A three-tiered evaluation index system for the integrated carrying capacity is constructed. The index system covers three dimensions: transportation subsystem, power subsystem, and integrated collaborative subsystem. The transportation subsystem index includes road network traffic efficiency and charging waiting time, and the data for these indexes are directly taken from the output results of SUMO traffic simulation. The power subsystem index includes voltage stability coefficient and line load rate, and the data for these indexes are obtained through distribution network power flow calculation using Matlab. The integrated collaborative subsystem index includes coupling coordination degree and charging request satisfaction rate, and the data for these indexes are calculated based on real-time interactive data between the two networks, thus achieving a comprehensive characterization of the carrying capacity status of the integrated network. Model accuracy verification and carrying capacity level classification were carried out: Traffic operation data, power distribution network operation data and charging service data collected from actual scenarios were used as verification benchmarks. The simulated values ​​of each dimension of indicators output by the joint simulation were compared with the actual benchmark data, and the relative error was calculated. The error control standard was set to not exceed 5% to ensure the accuracy of model evaluation. Based on the optimization results after verification, a five-level carrying capacity level classification standard of "excellent, good, medium, critical, and overloaded" was established to realize the quantitative classification of the carrying capacity status of the integrated network.

[0009] Further, in step S2, the TCP / IP protocol interaction is implemented as follows: The SUMO side constructs a TCP / IP server based on its natively supported TraCI (Traffic Control Interface) interface. The server is configured with a fixed communication port of 8080, which conforms to the commonly used port configuration specifications in the SUMO co-simulation field. A port listening mechanism is also enabled to respond to client connection requests in real time. The Matlab side initializes the TCP / IP client through its built-in tcpip function. The client configuration explicitly associates the server's IP address and the 8080 communication port. After completing the connection parameter configuration, a connection request is executed to establish an end-to-end communication link. At the data transmission level, JSON format is used for encapsulation. The encapsulated content includes data identifiers, timestamps, and core interactive data. The timestamp is used to achieve time synchronization between the two networks. This data encapsulation method conforms to the cross-software data interaction standards recommended by mainstream simulation platforms such as NSF-OAC, ensuring data parsing compatibility between SUMO and Matlab. Through the above configuration, the data transmission delay between the two networks is controlled within 1 second, meeting the real-time simulation requirements for data interaction in the transportation-power integration scenario.

[0010] Further, in step S4, the specific process of solving the strategy is as follows: First, a set of traffic OD demand samples is initialized as input parameters of the upper-level model. The samples are input to the lower-level co-simulation system. The operation status data of the traffic network and power network are obtained through dynamic coupling simulation of SUMO and Matlab. The constraint satisfaction is evaluated based on the preset constraints. Then, a surrogate model is constructed using Gaussian process regression of Bayesian optimization algorithm. New traffic OD demand samples are generated based on the expected improvement criterion. The above simulation and sample update process is repeated until the objective function of the upper-level model converges. The convergence criterion is that the fluctuation range of the calculated value of the objective function for 10 consecutive generations does not exceed 0.5%. At this time, the output objective function value is the optimal solution of the fusion network carrying capacity. Furthermore, in step S4, the parameter settings of the Bayesian optimization algorithm are adapted to the proposed simulation optimization method, specifically as follows: The Gaussian process regression module uses the radial basis function (RBF) as the kernel function, and constructs the covariance matrix between sample points through this kernel function to quantify sample similarity, matching the modeling requirements of the objective function for Gaussian process regression; the desired improvement EI function is selected as the acquisition function, and efficient optimization is achieved by balancing exploration (unknown region search) and utilization (known optimal region refinement) in the optimization process; the number of iterations is set to 50-100 times, and this parameter range has been experimentally verified to be adaptable to complex constraint scenarios of fusion networks, ensuring that the solution process has both efficiency and accuracy.

[0011] The beneficial effects of this application are as follows: (1) Implementing a real-time interaction mechanism between two networks supported by the TCP / IP protocol and verifying its feasibility: A bidirectional data interaction system was constructed based on SUMO's native TraCI interface and Matlab's built-in TCP / IP toolbox. The TraCI interface natively adapts to the TCP / IP protocol through the Socket communication mechanism, supporting real-time data reading and writing and control command transmission during the simulation process, which conforms to the cross-software interaction architecture recommended by the NSF-OAC simulation platform. The Matlab side completes client initialization through the tcpip function and configures communication parameters matching the TraCI server to achieve a stable connection. This architecture verifies the engineering feasibility of cross-software real-time data transmission, solves the problem of lack of collaboration caused by data splicing in traditional independent simulation, and makes the dynamic interaction characteristics of the two-network coupled simulation process highly consistent with the actual fusion scenario.

[0012] (2) Strong adaptability of model paradigm and solution method: The two-layer model paradigm of "upper-layer system optimization - lower-layer operation simulation" is introduced, which clearly defines the upper-layer objective of "maximizing the carrying capacity of the integrated network" and the lower-layer requirement of "satisfying the dynamic operation constraints of transportation and power", and adapts to the hierarchical optimization logic of the integrated scenario. The solution process adopts the Bayesian optimization simulation optimization method, and constructs a proxy model of the objective function through Gaussian process regression in order to improve the exploration and utilization of the function balance optimization process. Multi-stage penalty factors are introduced to handle second-order constraints such as travel time, charging service and carbon emissions. This retains the optimization kernel of the traditional analytical model and overcomes its defect of being unable to handle dynamic coupling constraints. Experiments have verified that the solution efficiency is more than 30% higher than that of the traditional analytical method.

[0013] (3) Comprehensive and accurate carrying capacity assessment: The theoretical framework of the traditional analytical model is integrated with the dynamic simulation capability of the SUMO-Matlab co-simulation. The upper-level model coordinates the carrying capacity maximization objective under second-order constraints such as travel time, carbon emissions, and charging service duration. The lower-level model transmits charging demand and power supply constraint signals in real time through the TraCI-TCP / IP interactive channel. Combined with the Logit charging decision algorithm, the charging behavior of electric vehicles and the dynamic coupling process of the two networks are accurately characterized. The case verification logic is adopted, and the accuracy verification is carried out based on actual traffic operation data, distribution network parameters, and charging service data. The relative error between the assessment results and the actual values ​​is ≤5%, realizing a comprehensive and accurate characterization of the carrying capacity status of the integrated network.

[0014] (4) Excellent engineering practicality: This method is based on two mature commercial software programs, SUMO and Matlab, to build the core simulation module, and relies on the TraCI interface and TCP / IP protocol to build the interactive architecture. The two-layer model paradigm can be adapted to different scale scenarios by adjusting parameters such as OD demand range and distribution network voltage level. The simulation optimization solution method can be directly embedded into the existing integrated network analysis platform. In practical applications, the output carrying capacity level results and key constraint bottleneck data can provide direct technical support for the selection of charging facility sites, capacity configuration and the formulation of peak-hour coordinated scheduling schemes for the two networks, significantly reducing the trial and error costs of engineering planning and operation optimization. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of the two-layer model framework and overall process of the present invention; Figure 2 This is a diagram illustrating the implementation path of SUMO-Matlab real-time interactive logic and co-simulation. Figure 3 This is the iterative convergence curve of the Bayesian optimization algorithm for solving a two-level model. Figure 4 This is a comparison chart of the integrated bearing capacity assessment results and the traditional bearing capacity in the embodiment. Detailed Implementation

[0016] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that the following specific embodiments are for illustrative purposes only and are not intended to limit the scope of the invention.

[0017] As shown in the figure, this embodiment provides a method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation. The scenario is the G15 Shenhai Expressway from Suzhou Shaxi Service Area to Nantong Xianfeng Service Area (approximately 85km in length) and its supporting power distribution network. The transportation network includes one two-way six-lane main line, four interchanges (Shaxi Interchange, Dongbang Hub, Xianfeng Interchange, etc.), and three service area charging stations (Shaxi Service Area North / South, Xianfeng Service Area North / South, for a total of six charging zones). The charging facilities adopt 120kW DC fast charging piles, with 8 piles and 16 guns configured in each zone. The rated power of a single pile is 120kW, and dynamic power allocation is supported. The OD demand is based on the monitoring data of Jiangsu Provincial Expressway Network, which is 8000-12000 veh / h on weekdays and 15000-20000 veh / h during peak holiday periods. Electric vehicles account for 18% on weekdays and 25% on holidays. The power distribution network adopts a 10kV radial power supply network, relying on three 110kV / 10kV substations along the line (Taicang Shaxi Substation, Changshu Dongbang Substation, and Nantong Xianfeng Substation). Each substation has a rated capacity of 100MVA and is equipped with two 10kV feeders specifically for supplying the charging load of the corresponding service area. The feeder cross-section is 240mm². The base load of the power distribution network along the line is 35MW (including the power load of the service area, and the load of surrounding residents and industries). The carbon emission limit is set at 8500kg / h with reference to the Jiangsu Province power industry standard. The average travel time threshold is set at 510s (85km / 100km / h design speed) with reference to the design standard of this section.

[0018] First, data collection and preprocessing are performed: Traffic data: Highway OD traffic data were obtained through the Jiangsu Provincial Expressway Network Operation and Management Center. The road network topology was extracted using OpenStreetMap and calibrated using SUMO's Netedit tool, with the calibration error controlled within 5%. Electric vehicle parameters were referenced from Jiangsu Province's new energy vehicle operation monitoring data, selecting mainstream models BYD Han EV and Tesla Model 3, with battery capacities of 85kWh and 60kWh respectively. The range degradation coefficient was set to 0.95, and the charging demand trigger threshold was set to ≤25% remaining battery power.

[0019] Power data: Parameters of three substations and feeders along the line were obtained from State Grid Jiangsu Electric Power Co., Ltd. The 10kV feeders use steel-cored aluminum stranded wire with a resistance of 0.18Ω / km and a reactance of 0.08Ω / km. The line lengths are 3.2km from Shaxi substation to Shaxi service area, 5.8km from Dongbang substation to intermediate charging point, and 4.5km from Xianfeng substation to Xianfeng service area. The main transformer ratio of the substations is 110kV / 10.5kV, and the short-circuit impedance is 6.5%. The baseline load data uses the average monitoring value of the same period in 2024, with a baseline load of 12MW for the Shaxi substation supply area, 15MW for Dongbang substation, and 8MW for Xianfeng substation. The above parameters were imported into Matlab Power System Toolbox, and a power flow calculation model was constructed using the Newton-Raphson method, with a convergence accuracy set to 10⁻. 6 .

[0020] Data coupling: A one-to-one mapping relationship is established between charging facilities and power distribution network feeders (Shaxi service area North / South zones correspond to Shaxi substation 10kV feeder 1# / 2#, intermediate charging points correspond to Dongbang substation 10kV feeder 3# / 4#, Xianfeng service area North / South zones correspond to Xianfeng substation 10kV feeder 5# / 6#); TCP / IP communication parameters are configured according to industrial-grade standards: SUMO side IP: 192.168.5.108, port 8080, using TraCI version 1.12 interface; Matlab side IP: 192.168.5.109, port 8081, data transmission is encapsulated in JSON format with added CRC-16 check bits, and the communication cycle is set to 3 minutes to meet real-time interaction requirements.

[0021] Secondly, the specific implementation steps are as follows: Step S1: The objective function of the upper-level model is set as (Q is the total traffic OD demand, η is the power supply matching coefficient, and γ is the charging service satisfaction coefficient), where γ = number of charging requests satisfied / total number of charging requests; the constraints are set according to the actual scenario: average travel time ≤ 510s, power grid carbon emissions ≤ 8500kg / h, single-pile charging service duration ≤ 45min (the conventional time for a 120kW fast charging pile to charge to 80%), distribution network node voltage deviation ≤ ±5%, and feeder load rate ≤ 85%. The lower-level model clarifies the traffic flow conservation constraints (segment inflow = outflow + dwell time), electric vehicle charging power constraints (0-120kW dynamically adjustable), and feeder current carrying capacity constraints (240mm² conductor current carrying capacity 310A).

[0022] Step S2: Start the SUMO TraCI server at the Taicang monitoring center and start the Matlab TCP / IP client at the Nantong power dispatch terminal. Verify link connectivity using the ping command, with a packet loss rate ≤0.1%. SUMO sends charging request data (format example: "Shaxi South - 10:05 - 120kW - Feeder 1#") and road network status data (including traffic flow and average vehicle speed for each lane) every 3 minutes. Matlab completes power flow calculation and sends back a signal within 10 seconds of receiving the data (format example: "Feeder 1# - Load rate 86% - Current limit to 100kW"). A continuous 72-hour stability test shows that the data transmission delay is ≤0.8s, with no data loss or errors, meeting the requirements for real-time interaction.

[0023] Step S3: Embed an improved Logit charging decision algorithm into SUMO, adding a weight factor for charging price (Jiangsu peak-valley electricity price: 1.05 yuan / kWh during peak hours and 0.38 yuan / kWh during off-peak hours) and charging pile idle status. When the remaining battery power of an electric vehicle is ≤25%, a charging request is triggered. The algorithm comprehensively calculates the utility value of each charging zone and selects the optimal station. During the 11:00-11:30 period on holidays, the demand for 120kW charging at the Shaxi South Service Area is concentrated. Matlab calculation shows that the load rate of feeder #1 reaches 87.2% (exceeding the 85% threshold), and a "current limit to 90kW + guidance diversion" signal is fed back. After SUMO responds, the charging power of this zone is reduced to 90kW, and 30% of the vehicles waiting to be charged are guided to a backup charging point 5km away through navigation instructions. After the adjustment, the feeder load rate drops to 79.5%, and the average vehicle speed on the road network is maintained at 92km / h.

[0024] Step S4: The Bayesian optimization algorithm is used to solve the two-layer model. The radial basis function is still used as the kernel function, and the improved expectation improvement function (introducing a constraint penalty term) is used as the acquisition function. Six sets of OD demand samples are initialized (8000, 10000, and 12000 veh / h on weekdays, and 15000, 18000, and 20000 veh / h on holidays), and the constraint satisfaction status is obtained by inputting them into the lower-level co-simulation system. A surrogate model is constructed through Gaussian process regression. After 42 iterations, the objective function converges (the fluctuation range is 0.32% ≤ 0.5% for 10 consecutive generations). The optimal solution corresponds to a holiday OD demand of 16500 veh / h. At this time, the peak electric vehicle charging load is 4.8MW, and the optimal value of the integrated carrying capacity is 16500 veh × 0.92 (power supply matching coefficient) × 0.94 (charging service satisfaction coefficient) = 14254 veh·MW.

[0025] Step S5: Based on the three-level evaluation index system, the core indicators are calculated as follows: average travel time 485s (≤510s), charging waiting time 12min, distribution network node voltage deviation 3.2% (≤±5%), average feeder load rate 76.8% (≤85%), coupling coordination degree 0.86, and charging request satisfaction rate 94%. The comprehensive score calculated using the analytic hierarchy process is 87 points, corresponding to a "good" carrying capacity level. Verification was performed using actual monitoring data from the 2024 National Day holiday. The relative errors between the simulated and actual values ​​were: traffic flow 2.8%, charging load 3.5%, and voltage deviation 2.1%, all ≤5%, meeting the accuracy requirements.

[0026] Finally, the implementation results are compared as follows: Comparing the method of this invention with traditional analytical methods and single simulation methods: Traditional analytical methods use static mathematical models to characterize carrying capacity, without considering the dynamic interaction between traffic flow and grid current and the pressure change process. This leads to a significant deviation between the carrying capacity calculation and the actual resource utilization state. The planned charging facility layout results in a road network traffic resource utilization rate of only 62% (with idle lanes). At the same time, the load rate of the distribution network transformer fluctuates too much (35%-92%), causing both power resource waste and local overload, and failing to achieve coordinated optimization of the two networks. The single SUMO traffic simulation only focuses on road network traffic efficiency, does not incorporate grid operation constraint feedback, and underestimates the centralized access of charging load. The simulation failed to regulate the charging load during peak hours, resulting in a voltage deviation of -8.2% at distribution network nodes (exceeding the ±5% safety threshold), triggering the grid protection device and causing an actual charging service interruption rate of 18%. Furthermore, the single Matlab distribution network simulation used a fixed charging load curve input, failing to consider the spatiotemporal mismatch of charging demand caused by traffic congestion, thus underestimating the operational pressure on the traffic network. The simulated peak charging load periods were out of sync with actual traffic congestion periods, resulting in a simulated charging wait time of only 22 minutes, while the actual waiting time due to vehicle congestion reached 58 minutes, exceeding the 30-minute service quality threshold. This invention clarifies the hierarchical relationship between system optimization objectives and operational constraints through a two-layer model. It combines real-time TCP / IP interaction to achieve dynamic coupling simulation of traffic flow and grid current. Simultaneously, it uses a Bayesian optimization algorithm to achieve coordinated matching of resources between the two networks. Ultimately, the utilization rate of road network traffic resources is increased to 85%, the voltage deviation of distribution network nodes is controlled within -2.5% (meeting the safety threshold), and the actual charging waiting time is shortened to 18 minutes. This not only avoids the risk of grid voltage exceeding limits but also alleviates traffic congestion, verifying the superiority of the method.

[0027] The technical means disclosed in this invention are not limited to those disclosed in the above embodiments, but also include technical solutions composed of any combination of the above technical features. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications are also considered within the scope of protection of this invention.

Claims

1. A method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation, characterized in that, Includes the following steps: Step S1: Construct a two-layer model framework that includes upper and lower layer models. The upper layer model aims to maximize the total carrying capacity of the fusion network. Its constraints include the average travel time threshold of the transportation network, the carbon emission limit of the power network, and the service duration limit of charging facilities. The lower-level model is the SUMO-Matlab co-simulation layer, which aims to satisfy the dynamic operation constraints of the traffic network and the power network. Its constraints include traffic flow conservation constraints, distribution network voltage deviation threshold control within ±5%, line load rate upper limit set at 85%, and electric vehicle charging power constraints. Step S2: Based on the TCP / IP protocol, a bidirectional data interaction channel is built using the TraCI interface built into SUMO and the TCP / IP toolbox integrated into Matlab to realize dynamic feedback of charging demand and power supply constraint signals; The TraCI interface natively supports the TCP / IP communication protocol, and the TCP / IP toolbox completes the configuration for data reception and transmission. Step S3: Construct a lower-level SUMO-Matlab co-simulation model, in which a traffic simulation model containing the Logit charging decision algorithm is constructed in the SUMO traffic simulation model, and a distribution network simulation model containing charging load is constructed in the Matlab distribution network simulation model, so as to realize the dynamic coupling simulation of the two networks. Step S4: Use Bayesian optimization algorithm to solve the two-layer model, construct surrogate model through Gaussian process regression, and iteratively optimize based on expectation improvement criterion to obtain the optimal solution for bearing capacity; Step S5: Establish a three-level evaluation index system, verify the model using actual data, and classify the bearing capacity levels.

2. The method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation as described in claim 1, characterized in that, In step S2, the TCP / IP protocol interaction is implemented as follows: SUMO builds a TCP / IP server based on its natively supported TraCI interface. The server is configured with a fixed communication port of 8080, which conforms to the common port configuration specifications in the SUMO co-simulation field. A port listening mechanism is also enabled to respond to client connection requests in real time. Matlab initializes the TCP / IP client using its built-in tcpip function. The client configuration explicitly associates the server's IP address and the 8080 communication port. After configuring the connection parameters, a connection request is executed to establish an end-to-end communication link. Data transmission is encapsulated in JSON format, containing data identifiers, timestamps, and core interactive data. The timestamps are used to synchronize the simulation times of the two networks. This configuration keeps the data transmission delay between the two networks within 1 second, meeting the real-time simulation requirements for data interaction in the transportation-power integration scenario.

3. The method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation as described in claim 1, characterized in that, In step S2, the interactive data format and transmission rules are defined: At the data sending end, SUMO collects and outputs two types of core data in real time. One type is electric vehicle charging request data, which specifically includes charging facility identification, charging power demand, and estimated charging time. The other type is road network operation status data, which specifically includes traffic flow of each road segment and average vehicle speed of the road network. The above data is encapsulated by the TCP / IP protocol and transmitted to Matlab. At the data receiving and feedback end, Matlab receives the data and performs power flow calculation of the distribution network in real time to obtain the operating status parameters of the distribution network node voltage and line load rate. The calculation results are compared with the preset constraint threshold. If the threshold is triggered, a power supply constraint signal is generated and fed back to SUMO through the same interactive channel. SUMO adjusts the charging facility operation parameters or the electric vehicle driving path according to the signal.

4. The method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation as described in claim 1, characterized in that, In step S3, the SUMO traffic simulation model imports the road network structure, OD matrix, and electric vehicle parameters, and embeds a charging decision algorithm. This charging decision algorithm uses the Logit model, and during the model construction process, it comprehensively considers the distance between charging facilities and vehicles, as well as the time preference of vehicles traveling to charging facilities. When receiving power supply constraint signals from Matlab, the model responds in real time and executes adjustment strategies, specifically including dynamically reducing the operating power of charging facilities or guiding vehicles waiting to be charged to plan their travel paths to other compliant charging facilities, thereby achieving dynamic matching between traffic operation status and power constraints. The Matlab distribution network simulation model imports the distribution network topology, line parameters, and baseline load data. After receiving the electric vehicle charging demand data output by SUMO through the TCP / IP interactive channel, it uses this data as a dynamic load to access the distribution network simulation scenario. The Newton-Raphson method is used to perform real-time power flow calculations to obtain the core operating parameters of distribution network node voltage and line load rate, which are then compared with preset power constraint thresholds to complete compliance judgment.

5. The method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation as described in claim 1, characterized in that, In step S4, the specific process of solving the strategy is as follows: First, a set of traffic OD demand samples is initialized as input parameters of the upper-level model. The samples are input to the lower-level co-simulation system. The operation status data of the traffic network and power network are obtained through dynamic coupling simulation of SUMO and Matlab. The constraint satisfaction is evaluated based on the preset constraints. Then, a surrogate model is constructed using Gaussian process regression of Bayesian optimization algorithm. New traffic OD demand samples are generated based on the expected improvement criterion. The above simulation and sample update process is repeated until the objective function of the upper-level model converges. The convergence criterion is that the fluctuation range of the calculated value of the objective function for 10 consecutive generations does not exceed 0.5%. At this time, the output objective function value is the optimal solution of the fusion network carrying capacity.

6. The method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation as described in claim 1, characterized in that, In step S4, the parameter settings of the Bayesian optimization algorithm are adapted to the proposed simulation optimization method, as follows: The Gaussian process regression module uses the radial basis function RBF as the kernel function, and constructs the covariance matrix between sample points through the kernel function to quantify sample similarity, matching the modeling requirements of the objective function for Gaussian process regression; the desired improvement EI function is selected as the acquisition function, and efficient optimization is achieved by balancing exploration and utilization in the optimization process; the number of iterations is set to 50-100 times.

7. The method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation as described in claim 1, characterized in that, In step S5, a three-level integrated carrying capacity assessment index system is constructed: the index system covers three dimensions: traffic subsystem, power subsystem and integrated collaborative subsystem. The traffic subsystem indexes include road network traffic efficiency and charging waiting time. The data of these indexes are directly taken from the output results of SUMO traffic simulation. The power subsystem indexes include voltage stability coefficient and line load rate. The data of these indexes are obtained through Matlab distribution network power flow calculation. The indicators of the converged and coordinated subsystem include coupling coordination degree and charging request satisfaction rate. These indicators are calculated based on real-time interactive data between the two networks, achieving a comprehensive characterization of the carrying status of the converged network.

8. The method for analyzing the carrying capacity of a transportation-electricity integrated network based on two-layer co-simulation as described in claim 1, characterized in that, In step S5, model accuracy verification and carrying capacity level classification are carried out: traffic operation data, power distribution network operation data and charging service data collected in actual scenarios are used as verification benchmarks. The simulation values ​​of each dimension of indicators output by the joint simulation are compared with the actual benchmark data, and the relative error is calculated. The error control standard is set to not exceed 5% to ensure the accuracy of model evaluation. Based on the optimization results after verification, a five-level carrying capacity level classification standard of "excellent, good, medium, critical, and overloaded" is established to realize the quantitative classification of the carrying capacity status of the integrated network.

Images