New engineering traffic impact assessment simulation prediction method

Abstract

This invention relates to the fields of intelligent transportation systems and urban planning technology, specifically a novel simulation and prediction method for engineering traffic impact assessment. First, it constructs a multi-source dynamic data framework integrating mobile phone signaling, roadside sensing, weather, and event data. Second, it innovatively employs a three-layer hybrid model of "GWRFR+LSTM+Kalman filtering" to sequentially capture the spatial heterogeneity and nonlinear relationship between traffic and the environment, learn the spatiotemporal dependence characteristics of traffic flow, and perform real-time corrections for sudden disturbances. Then, it divides the dynamic parameter set for the entire lifecycle of the project, introduces uncertainty factors, and outputs probabilistic evaluation results through Monte Carlo simulation. Finally, it quantifies the interactive impact of modes such as private cars, public transportation, and cycling through a multi-modal traffic coordination module. This invention significantly improves prediction accuracy and dynamic adaptability, providing more scientific and reliable support for the comparison of engineering schemes and traffic management decisions.

Description

Technical Field

[0001] This invention relates to the fields of intelligent transportation systems and urban planning technology, specifically a novel simulation and prediction method for engineering traffic impact assessment. Background Technology

[0002] Traffic Impact Assessment (TIA) is a key technical method for evaluating the impact of new or expanded construction projects on the traffic operation of the surrounding road network, and it serves as an important basis for project planning and decision-making. Traditional TIA methods have the following limitations: Static data lag: Over-reliance on static data such as historical traffic surveys and land use planning makes it difficult to reflect real-time, dynamic traffic conditions and emergencies.

[0003] The model is simple and crude: it often uses the four-stage method based on static traffic assignment, with fixed model parameters, which makes it difficult to capture the complex spatiotemporal nonlinear relationships, spatial heterogeneity, and interactive effects between multiple traffic modes in traffic flow.

[0004] The evaluation dimensions are too narrow: they focus mainly on predicting motor vehicle traffic flow, and lack collaborative analysis and quantitative evaluation of multi-modal transportation systems such as public transportation and slow-moving transportation.

[0005] Lack of consideration for uncertainties: Insufficient consideration of real uncertainties such as construction delays and fluctuations in passenger flow after operation; evaluation results are mostly deterministic outputs; weak risk warning capabilities.

[0006] Poor dynamic adaptability: It is difficult to continuously and dynamically track, evaluate, predict and correct the entire life cycle of an engineering project from the construction phase to the operation phase.

[0007] Therefore, there is an urgent need for a new simulation and prediction method for engineering traffic impact assessment that can integrate real-time dynamic data, handle complex spatiotemporal relationships, quantify uncertainties, and support multi-mode collaborative analysis, in order to overcome the shortcomings in current practical applications. Summary of the Invention

[0008] The purpose of this invention is to provide a novel simulation and prediction method for engineering traffic impact assessment, in order to solve the problems mentioned in the background art.

[0009] To achieve the above objectives, the present invention provides the following technical solution: A novel simulation and prediction method for engineering traffic impact assessment includes the following steps: S1. Construct a multi-source heterogeneous dynamic data fusion framework to access and integrate mobile phone signaling data, roadside sensor data, weather and event data, and engineering project feature data; S2. Based on the fused data, a hybrid prediction model of "GWRFR+LSTM+Kalman filter" is used to dynamically predict traffic flow, wherein: S2.1. The spatial heterogeneity and nonlinear relationship between built environment variables and traffic flow are captured by the geographically weighted random forest regression model (GWRFR) to generate the basic traffic demand distribution. S2.2 Input the output of GWRFR and real-time time series data into a Long Short-Term Memory (LSTM) network to learn the spatiotemporal dependence characteristics of traffic flow and obtain medium- and short-term prediction values; S2.3 Using the Kalman filter algorithm, the LSTM prediction value is used as the state prediction, and the real-time sensor data is used as the observation value to correct the prediction result in real time, so as to obtain the final disturbance-resistant dynamic traffic flow prediction result. S3. Define the dynamic parameter set for the construction and operation phases of the project, identify key uncertainty factors such as construction delays and passenger flow fluctuations, and use the Monte Carlo simulation method to run the hybrid prediction model multiple times to output the probability distribution of traffic impact indicators, thereby realizing the probabilistic evaluation results. S4. Establish a multi-modal traffic coordination analysis module to quantify the interaction and mode shift among private cars, public transportation, and slow traffic, and assess the impact of different traffic organization schemes on the overall road network.

[0010] As a further aspect of the present invention: in S1, the mobile phone signaling data is used to extract the dynamic population distribution and travel trajectory matrix (OD matrix); the roadside sensor data includes real-time traffic flow, speed, and occupancy information collected by coils, cameras, or radar; the weather and event data includes weather warnings and forecasts of large-scale events; the project feature data includes construction plans, construction schedules, post-construction business attributes, and predicted passenger flow.

[0011] As a further aspect of the present invention: In S2.1, the GWRFR model constructs a localized random forest model for each traffic analysis cell or road segment. The input variables of the model include built environment variables such as land use density, road network density, and bus stop density, as well as historical traffic flow data.

[0012] As a further aspect of the present invention: in S2.2, the time-series data input to the LSTM also includes weekday type, holiday identifier, and real-time weather information.

[0013] As a further aspect of the present invention: in S2.3, the Kalman filter treats sudden traffic accidents and short-term extreme weather as system noise and observation noise for processing and correction.

[0014] As a further aspect of the present invention: In S3, the Monte Carlo simulation takes the probability distribution of uncertainty factors as input, conducts tens of thousands of simulation experiments, and outputs the percentile results of service levels and delay times of key road segments or nodes.

[0015] As a further aspect of the present invention: In S4, the multi-modal traffic cooperative analysis module establishes a mode division and transfer model to quantify the impact of adjusting bus lane settings, parking fee policies, or cycling environment improvement measures on the overall road network traffic load.

[0016] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, implements the novel engineering traffic impact assessment simulation and prediction method as described in any of the preceding claims.

[0017] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the novel engineering traffic impact assessment simulation and prediction method as described in any of the preceding claims.

[0018] Compared with the prior art, the beneficial effects of the present invention are: Significantly improved prediction accuracy: Through multi-source dynamic data fusion and a three-layer hybrid model, complex traffic systems can be characterized more accurately, and the prediction error can be reduced by more than 30% compared with traditional methods; Strong dynamic and adaptive capabilities: The framework can access real-time data streams, and the Kalman filter layer can quickly respond to sudden disturbances, enabling dynamic tracking and real-time correction of the impact on engineering projects. More scientific evaluation results and visible risks: By outputting probabilistic results through Monte Carlo simulation, decision-makers are provided with the range of impact under different confidence levels, which significantly enhances risk assessment and response capabilities; Comprehensive analysis dimensions: For the first time, multi-modal transportation cooperative interaction is systematically incorporated into the quantitative evaluation framework, supporting the selection of more comprehensive and sustainable transportation solutions; Full lifecycle coverage: The methodology spans both the construction and operation phases of an engineering project, enabling closed-loop management from planning to post-evaluation and supporting continuous optimization; It possesses strong patent innovation and practicality: The proposed technical route of "data fusion framework + three-layer hybrid model + uncertainty quantification + multi-mode collaboration" integrates cutting-edge technologies from multiple disciplines such as geographic information science, machine learning, signal processing, and traffic engineering, solves long-standing pain points in the industry, conforms to the development trend of intelligent transportation, and has high feasibility and application value. Attached Figure Description

[0019] Figure 1 This is a flowchart of a new method for simulating and predicting the impact of engineering traffic on transportation. Detailed Implementation

[0020] The technical solution of this application will be further described in detail below with reference to specific embodiments.

[0021] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0022] Please see Figure 1 In one embodiment of the present invention, a novel engineering traffic impact assessment simulation and prediction method is proposed. Its core lies in constructing a "multi-source dynamic data fusion traffic impact assessment framework" and innovatively employing a hybrid model of "GWRFR + LSTM + Kalman filtering" for dynamic simulation and prediction. The method specifically includes the following steps: Step S1: Construct a multi-source heterogeneous dynamic data fusion framework, establish a unified data access and processing platform, and integrate and fuse the following multi-source heterogeneous data in real-time / near real-time: Dynamic population flow and OD data: mobile signaling data used to extract dynamic population distribution and travel trajectories (OD matrix) at a fine spatiotemporal level.

[0023] Real-time traffic flow data: roadside sensors (such as loop detectors, cameras, and radar) and floating car GPS data provide real-time information on road network speed, traffic flow, and occupancy.

[0024] Environmental and event data: real-time weather and warning information from meteorological departments (such as rainfall, snow), forecasts of major events, and information on social media events.

[0025] Project characteristic data includes construction plan, construction period plan, traffic organization plan, post-completion project attributes (such as commercial area, office area, parking spaces, etc.) and predicted passenger flow during operation.

[0026] Step S2: Construct a three-layer hybrid prediction model of "GWRFR + LSTM + Kalman filter". S2.1, GWRFR (Geographically Weighted Random Forest Regression) Layer - Modeling Spatial Heterogeneity and Nonlinear Relationships: Input: Built environment variables (such as land use density, road network density, bus stop density) and historical / base year traffic flow data.

[0027] Processing: Using the idea of ​​Geographically Weighted Regression (GWR), a localized Random Forest (RF) model is constructed for each traffic analysis zone (TAZ) or road segment, thereby simultaneously capturing the complex nonlinear relationship between traffic generation and attraction and the built environment, as well as its spatial heterogeneity (i.e., the same factor has different influences in different regions).

[0028] Output: High-precision basic traffic demand distribution (spatialized flow and OD prior estimates).

[0029] S2.2, LSTM (Long Short-Term Memory) Layer - Spatiotemporal Dependency Feature Learning: Input: Basic traffic flow output from the GWRFR layer, which integrates real-time traffic flow, time characteristics (weekdays, holidays), weather conditions and other time-series data.

[0030] Processing: Construct graph neural networks or sequence LSTM networks based on road network topology to learn the complex dependencies and propagation patterns of traffic flow in time and space (such as congestion propagation and morning / evening peak patterns).

[0031] Output: Short-term traffic flow forecasts considering spatiotemporal dependencies.

[0032] S2.3, Kalman Filter Layer - Real-time Correction and Burst Disturbance Handling: Input: The predicted values ​​of the LSTM layer are used as state predictions, and real-time sensor data are used as observations.

[0033] Processing: Disturbances such as sudden traffic accidents and short-term heavy rainfall that were not fully learned in historical models are treated as system noise and observation noise. The Kalman filter algorithm is used to perform real-time, recursive optimal estimation correction on the LSTM prediction results.

[0034] Output: Real-time calibrated final traffic flow dynamic prediction results with strong anti-disturbance capabilities.

[0035] Step S3: Full lifecycle dynamic parameter set and uncertainty quantification For the construction and operation phases, define dynamic parameter sets (such as the scope and duration of construction enclosures, frequency of heavy vehicle entry and exit, project-induced passenger flow, mode division rate, etc.).

[0036] Identify key uncertainty factors, including the probability distribution of construction delays, the range of passenger flow fluctuations during operation, and the uncertainty of key model parameters (such as the V / C ratio threshold).

[0037] The Monte Carlo simulation method is used, with the above-mentioned uncertainty factors as random variables as inputs. The mixture model in step two is run thousands of times to obtain the probability distribution (such as P50 and P85 values) of traffic impact indicators (such as road segment service level and critical node delays). The probabilistic evaluation results are output to replace the traditional single deterministic values.

[0038] Step S4: Quantitative Analysis of the Impact of Multimodal Traffic Cooperation A multi-modal traffic collaboration analysis module has been added to establish an interaction and influence model among modes of transportation such as private cars, buses, and bicycles / cycling (e.g., the competitive relationship between cars and buses, and the substitution of cycling for short-distance motor vehicle travel).

[0039] Quantify the changes in traffic flow transfer and overall road network operating efficiency under different traffic organization schemes (such as adding dedicated bus lanes and optimizing the cycling environment) to solve the bias of traditional single-mode prediction.

[0040] This novel simulation and prediction method for engineering traffic impact assessment, through multi-source dynamic data fusion and a three-layer hybrid model, can more accurately characterize complex traffic systems, reducing prediction errors by more than 30% compared to traditional methods. The framework can access real-time data streams, and the Kalman filter layer can quickly respond to sudden disturbances, enabling dynamic tracking and real-time correction of engineering impacts. By outputting probabilistic results through Monte Carlo simulation, it provides decision-makers with the impact range at different confidence levels, significantly enhancing risk assessment and response capabilities. For the first time, it systematically incorporates multi-modal traffic collaboration into the quantitative evaluation framework, supporting the selection of more comprehensive and sustainable traffic solutions. The method spans both the construction and operation phases of engineering projects, achieving closed-loop management from planning to post-evaluation and supporting continuous optimization. The proposed technical route of "data fusion framework + three-layer hybrid model + uncertainty quantification + multi-modal collaboration" integrates cutting-edge technologies from multiple disciplines such as geographic information science, machine learning, signal processing, and traffic engineering, solving long-standing pain points in the industry, aligning with the development trend of intelligent transportation, and possessing high feasibility and application value.

[0041] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the concept of the present invention, and these should also be considered within the scope of protection of the present invention. These will not affect the effectiveness of the implementation of the present invention or the practicality of the patent.

Claims

1. A novel simulation and prediction method for engineering traffic impact assessment, characterized in that, Includes the following steps: S1. Construct a multi-source heterogeneous dynamic data fusion framework to access and integrate mobile phone signaling data, roadside sensor data, weather and event data, and engineering project feature data; S2. Based on the fused data, a hybrid prediction model of "GWRFR+LSTM+Kalman filter" is used to dynamically predict traffic flow, wherein: S2.

1. The spatial heterogeneity and nonlinear relationship between built environment variables and traffic flow are captured by the geographically weighted random forest regression model (GWRFR) to generate the basic traffic demand distribution. S2.2 Input the output of GWRFR and real-time time series data into a Long Short-Term Memory (LSTM) network to learn the spatiotemporal dependence characteristics of traffic flow and obtain medium- and short-term prediction values; S2.3 Using the Kalman filter algorithm, the LSTM prediction value is used as the state prediction, and the real-time sensor data is used as the observation value to correct the prediction result in real time, so as to obtain the final disturbance-resistant dynamic traffic flow prediction result. S3. Define the dynamic parameter set for the construction and operation phases of the project, identify key uncertainty factors such as construction delays and passenger flow fluctuations, and use the Monte Carlo simulation method to run the hybrid prediction model multiple times to output the probability distribution of traffic impact indicators, thereby realizing the probabilistic evaluation results. S4. Establish a multi-modal traffic coordination analysis module to quantify the interaction and mode shift among private cars, public transportation, and slow traffic, and assess the impact of different traffic organization schemes on the overall road network.

2. The novel engineering traffic impact analysis simulation prediction method according to claim 1, wherein, In step S1, the mobile phone signaling data is used to extract the dynamic population distribution and travel trajectory matrix (OD matrix); the roadside sensor data includes real-time traffic flow, speed, and occupancy information collected by coils, cameras, or radar; the weather and event data includes weather warnings and forecasts of large-scale events; and the project feature data includes construction plans, construction schedules, post-completion business attributes, and predicted passenger flow.

3. The novel engineering traffic impact assessment simulation prediction method according to claim 2, characterized in that, In S2.1, the GWRFR model constructs a localized random forest model for each traffic analysis cell or road segment. The input variables of the model include built environment variables such as land use density, road network density, and bus stop density, as well as historical traffic flow data.

4. The novel engineering traffic impact assessment simulation prediction method according to claim 3, characterized in that, In step S2.2, the time-series data input to the LSTM also includes weekday type, holiday identifier, and real-time weather information.

5. The novel engineering traffic impact assessment simulation prediction method according to claim 4, characterized in that, In S2.3, the Kalman filter treats sudden traffic accidents and short-term extreme weather as system noise and observation noise for processing and correction.

6. The novel engineering traffic impact assessment simulation and prediction method according to claim 1, characterized in that, In S3, the Monte Carlo simulation takes the probability distribution of uncertainty factors as input, conducts tens of thousands of simulation experiments, and outputs the percentile results of service levels and delay times for key road segments or nodes.

7. The novel engineering traffic impact assessment simulation and prediction method according to claim 2, characterized in that, In S4, the multi-modal traffic collaborative analysis module establishes a mode division and transfer model to quantify the impact of adjusting bus lane settings, parking fee policies, or cycling environment improvement measures on the overall road network traffic load.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the novel engineering traffic impact assessment simulation and prediction method as described in any one of claims 1 to 7.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the novel engineering traffic impact assessment simulation and prediction method as described in any one of claims 1 to 7.

Images