Virtual-real interaction and closed-loop driving parallel traffic system implementation method

Abstract

The invention discloses a virtual-real interaction and closed-loop driving parallel traffic system implementation method. The method comprises the following steps: dividing multiple parallel traffic spaces and constructing virtual-real space traffic parameter vectors; constructing an interaction model of multiple parallel traffic spaces; modeling a state equation and an observation equation of a multi-parallel virtual-real space; iterative operation of the parallel traffic system is realized through closed-loop logic of'perception-evaluation-diagnosis-optimization-implementation '. According to the invention, intelligent and precise management of the traffic system can be realized, so that the adaptability and robustness of the traffic system are improved, resource utilization is optimized, user experience is improved, energy consumption and exhaust emission are reduced, and the traffic industry is promoted to develop towards a more intelligent, efficient and sustainable direction.

Description

Technical Field

[0001] The present invention belongs to the field of urban road intelligent transportation technology, and in particular to a method for realizing a parallel transportation system with virtual-reality interaction and closed-loop drive. Background Art

[0002] With the acceleration of urbanization and the continuous growth of traffic demand, the transportation system faces many severe challenges. The number of vehicles and pedestrians continues to rise, and the randomness and unpredictability of their behavior make the traffic situation extremely complicated. Especially at key traffic nodes such as intersections, the frequent confluence, diversion and intersection of traffic flows make traffic congestion and delays increasingly prominent. Traditional traffic management methods have exposed many limitations in dealing with these complex and changeable traffic conditions, and it is difficult to meet the needs of efficient, intelligent and safe operation of modern transportation. An innovative solution is urgently needed to improve the current status of traffic management.

[0003] Through domestic and foreign literature searches, a patent document related to the present invention, "A Method for Constructing a Tunnel Parallel Traffic System" (application number 202210602542.0), was found. This patent mainly focuses on the problem of tunnel traffic flow data simulation and proposes a method for constructing a tunnel parallel traffic system. The principle is to match the vehicle data at the checkpoint based on the license plate number and calculate the vehicle driving status, calculate the traffic flow characteristic parameters and identify traffic abnormalities, and use the SUMO simulation software as a basis to simulate the real-time traffic flow status of the tunnel and reproduce the historical traffic flow operation status during the query time period. In this way, dynamic monitoring of tunnel traffic safety can be achieved to a certain extent. However, this method is only a technical solution provided for the specific scenario of the tunnel, and does not fully consider the cause diagnosis of traffic problems, does not clearly indicate the ability of closed-loop control, and lacks comprehensive analysis and response capabilities for the traffic system.

[0004] In addition to the problems reflected in the above patents, traditional traffic management methods generally have defects in spatial modeling and system logic. In terms of spatial modeling, most existing methods do not make a detailed and comprehensive division of traffic space. They often simply distinguish between road types, such as urban main roads, secondary roads, etc., without in-depth consideration of the dynamic changes in the degree of traffic congestion, the real-time impact of weather factors on traffic, and traffic differences under different functional requirements. This makes it impossible to accurately simulate and regulate different traffic scenarios in actual traffic management. For example, in bad weather, the traffic model cannot be adjusted in time according to weather changes, resulting in a decrease in road capacity and an increase in the risk of traffic accidents. In terms of system logic, traditional methods lack an effective closed-loop feedback mechanism. Most of them adopt an open-loop control method and do not fully consider iterative optimization based on problem diagnosis. This method cannot evaluate and optimize the operating status of the traffic system in real time, and cannot timely discover and solve traffic problems, such as data deviations caused by sensor failures, traffic models that do not match the actual situation, etc., which ultimately leads to increasingly serious problems such as increased traffic congestion, energy waste and environmental pollution. Summary of the invention

[0005] The purpose of the present invention is to address the problems existing in the above-mentioned prior art and propose a method for realizing a parallel transportation system with virtual-reality interaction and closed-loop drive. By constructing multiple parallel space models, the state equations and observation equations of the virtual and real spaces are accurately modeled, and effectively solved in the "perception-assessment-diagnosis-optimization-implementation" closed-loop logic, the intelligent and precise management of the transportation system is realized, so as to improve the adaptability and robustness of the transportation system, optimize resource utilization, enhance user experience, and reduce energy consumption and exhaust emissions at the same time, so as to promote the development of the transportation industry in a more intelligent, efficient and sustainable direction.

[0006] The technical solution to achieve the purpose of the present invention is to provide a method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system, the method comprising the following steps:

[0007] Step 1, divide multiple parallel traffic spaces and construct virtual and real space traffic parameter vectors;

[0008] Step 2: construct an interactive model of multiple parallel traffic spaces;

[0009] Step 3, modeling state equations and observation equations of multiple parallel virtual and real spaces;

[0010] Step 4: Realize the iterative operation of the parallel traffic system through the closed-loop logic of “perception-assessment-diagnosis-optimization-implementation”.

[0011] Furthermore, the multiple parallel traffic spaces are divided in step 1, specifically including:

[0012] Based on traffic flow, speed, vehicle density, weather conditions, road type and functional requirements, the traffic system is divided into a physical real space Ω r and multiple parallel digital virtual spaces n is the number of virtual spaces;

[0013] The traffic system division includes dividing into free space according to the degree of traffic congestion Lightly congested space Severely congested space

[0014] The traffic system division also includes clear sky space according to weather conditions Rainy Day Space Snow Space

[0015] The traffic system division also includes intelligent driving guidance function space according to functional requirements Functional space for transportation support of major events

[0016] Furthermore, step 2 of constructing an interactive model of multiple parallel traffic spaces specifically includes:

[0017] Definition from A i Virtual Space To A j Virtual Space The conversion function

[0018]

[0019] Among them, ω1, ω2, ω3 and ω4 are weight coefficients used to adjust the importance of each parameter in the conversion decision, where, is the real space Ω P The traffic parameter vector at time t, It is virtual space The traffic parameter vector at time t, is the control quantity of real space at time t; It is virtual space The simulated control quantity is generated according to the virtual space model and control strategy;

[0020] The interaction methods between different parallel spaces include:

[0021] (1) Data exchange: Data is exchanged between different parallel traffic spaces based on demand;

[0022] (2) Spatial transformation: When the transformation function satisfies , spatial transformation is performed, and the transformation process involves the adjustment of the perception layer, data processing layer, control execution layer and model management layer; where θ ij is the conversion threshold;

[0023] (3) Collaborative iteration: Multiple parallel traffic spaces work together to perform iterative optimization.

[0024] Furthermore, the state equations and observation equations of modeling multiple parallel virtual and real spaces described in step 3 specifically include:

[0025] Step 3-1, establish the state equation of the virtual space at time t:

[0026]

[0027] In the formula, is the virtual space at time t state vector, is the virtual space at time t Simulation credibility index, used to measure the reliability of virtual space simulation results; is the state transfer function based on the preset model at time t, Virtual Space Current Status and observational data Map to the state of the next moment; is the process noise at time t;

[0028] Step 3-2, establish the observation equation of virtual space at time t:

[0029]

[0030] In the formula, is the observation vector, They correspond to the observed values ​​of traffic flow, vehicle speed, and traffic environment complexity respectively; is the observation function, which is used to transform the virtual space state vector Mapping to observation space; is the observation noise;

[0031] Step 3-3, observation equation of real space at time t:

[0032]

[0033] In the formula, is the real space observation vector, They correspond to the observed values ​​of traffic flow, vehicle speed, and traffic environment complexity, respectively. is the real space observation function; is the real-space observation noise.

[0034] Furthermore, the iterative operation of the parallel traffic system is realized through the closed-loop logic of "perception-assessment-diagnosis-optimization-implementation" described in step 4, which specifically includes:

[0035] Step 4-1, perception phase: collect and fuse real space and virtual space data to provide a data basis for subsequent operations, while monitoring information related to interactions between different parallel spaces;

[0036] Step 4-2, evaluation phase: Calculate multiple evaluation indicators and conduct spatial evaluation to determine the virtual space simulation effect, and evaluate the impact of different parallel space interactions on the transportation system;

[0037] Step 4-3, diagnosis phase: trigger diagnosis based on specific thresholds to analyze the source of virtual space or real space problems;

[0038] Step 4-4, optimization phase: optimize virtual space parameters and real space control strategy according to the diagnosis results;

[0039] Step 4-5, implementation phase: Apply optimization strategies, update the real and virtual space states and observation vectors, prepare for the next round of closed loop, and ensure the effective execution of interactions between different parallel spaces.

[0040] Compared with the prior art, the present invention has the following significant advantages:

[0041] (1) Multi-parallel space modeling improves adaptability and flexibility: The present invention divides the traffic system into physical real space and multiple parallel digital virtual spaces based on a variety of factors, such as the degree of traffic congestion, weather conditions, functional requirements, etc. This modeling method enables the system to accurately simulate different traffic scenarios, whether it is normal traffic flow, traffic congestion, severe weather or major event scenarios, and can perform targeted analysis and management in the corresponding space. For example, in a traffic congestion space, you can focus on changes in congestion-related parameters and adjust strategies in a timely manner; in a severe weather space, you can fully consider the impact of weather factors on traffic. Compared with traditional traffic management methods, the system of the present invention has stronger adaptability and can quickly switch to a suitable spatial model for processing according to actual traffic conditions, thereby improving the flexibility and pertinence of traffic management and effectively responding to complex and changing traffic environments.

[0042] (2) Interaction between virtual and real space to achieve precise management: By constructing the virtual and real space state equation and observation equation, the interaction between virtual space simulation data and real space observation data is realized. The virtual space generates simulation values ​​based on the preset model, and continuously optimizes its own state in combination with some real space observation data. At the same time, the real space obtains feedback information from the virtual space through the observation equation. In the data fusion stage, the sensor accuracy, virtual space simulation credibility and traffic environment dynamic adjustment weight are comprehensively considered to make the fused data more accurate and reliable. This enables the system to grasp the traffic status more accurately. For example, the various indicators calculated in the evaluation stage (traffic congestion index, average travel time, virtual and real space fitting degree index, etc.) are closer to the actual traffic situation, thereby providing a more scientific basis for the formulation of traffic control strategies and improving the accuracy of traffic management, which is better than the single data source and rough estimation of traditional methods.

[0043] (3) Closed-loop logic optimizes system performance: The closed-loop logic of "perception-evaluation-diagnosis-optimization-implementation" enables the system to continuously monitor traffic conditions, identify problems in a timely manner and make adjustments. In the perception stage, real and virtual space data are fully collected and integrated to provide a rich and accurate data basis for subsequent operations. In the evaluation stage, multiple indicators are calculated and spatial evaluation is performed to fully understand the operating status of the traffic system and the simulation effect of the virtual space, as well as the interaction between different parallel spaces. In the diagnosis stage, the dynamic threshold trigger mechanism is used to accurately locate the source of the problem, whether it is an inaccurate virtual space model or interference in the real space. In the optimization stage, the virtual space parameters and real space control strategies are optimized according to the diagnosis results. For example, the virtual space uses a gradient algorithm to optimize parameters, and the real space adjusts traffic lights and speed limit instructions according to the virtual space optimization results. In the implementation stage, the optimization strategy is applied to the real traffic system and the system status is updated to prepare for the next round of closed loop. The entire closed-loop process is continuously circulated, allowing the system to continuously optimize performance according to changes in the traffic environment. Compared with traditional traffic management that lacks a dynamic adjustment mechanism, the system of the present invention can better adapt to dynamic changes in traffic and improve traffic efficiency.

[0044] (4) Parallel space interaction enhances synergy: The interaction mode between different parallel spaces (space conversion, data exchange, collaborative iteration) enhances the synergy of the system. Space conversion can be converted in different spaces (such as sunny space to rainy space, intelligent driving guidance space to major event traffic guarantee space, etc.) according to the conversion function and threshold. The conversion process involves adjustments at all levels to ensure that the system operates efficiently in different scenarios. Data exchange enables different spaces to share data, such as the intelligent driving guidance space exchanging data with the traffic congestion space to optimize the guidance route, thereby improving information utilization. Collaborative iteration allows multiple parallel spaces to work together, such as multiple related spaces in traffic congestion management jointly analyzing the causes and optimizing strategies. Compared with the relatively independent work of each part of traditional traffic management, the system of the present invention can achieve multi-space collaborative optimization and improve the ability to deal with complex traffic management issues.

[0045] The present invention is further described in detail below in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a system architecture diagram of the present invention.

[0047] Figure 2 It is a system flow chart of the present invention.

[0048] Figure 3 The present invention is a flowchart of a method verification in one embodiment. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0050] It should be noted that if there are descriptions involving "first", "second", etc. in the embodiments of the present invention, the descriptions of "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or suggesting their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0051] In one embodiment, in combination Figure 1 and Figure 2 The present invention provides a method for realizing a virtual-real interaction and closed-loop driven parallel traffic system, which specifically includes:

[0052] Step 1, divide multiple parallel traffic spaces and construct virtual and real space traffic parameter vectors;

[0053] Step 2: construct an interactive model of multiple parallel spaces;

[0054] Step 3, modeling state equations and observation equations of multiple parallel virtual and real spaces;

[0055] Step 4: Realize the iterative operation of the parallel traffic system through the closed-loop logic of “perception-assessment-diagnosis-optimization-implementation”.

[0056] Furthermore, in one embodiment, dividing the multiple parallel traffic spaces in step 1 specifically includes:

[0057] Step 1.1, multi-parallel traffic space division: Based on factors such as traffic flow, vehicle speed, vehicle density, weather conditions, road type, and functional requirements, the traffic system is divided into a physical real space Ω R and multiple parallel digital virtual spaces

[0058] The traffic system division includes dividing into free space according to the degree of traffic congestion Lightly congested space Severely congested space In the unobstructed space, traffic flows smoothly, vehicle movement is basically unobstructed, vehicle speeds remain at normal levels, traffic flow is moderate, and vehicle density is low; in the lightly congested space, traffic flow increases, vehicle speeds drop slightly, the distance between vehicles begins to narrow, and driving speed is affected to a certain extent; in the heavily congested space, traffic flow is large, vehicle speeds are significantly reduced, vehicle queuing is serious, vehicle density is high, and traffic moves slowly.

[0059] The traffic system division also includes clear sky space according to weather conditions Rainy Day Space Snow Space On clear days, the weather is good, the road friction is normal, the visibility is high, the impact on traffic is small, the vehicle speed is stable, and the traffic flow is normal; on rainy days, rainfall will cause the road surface to be slippery, the vehicle braking distance will increase, visibility will decrease, the driver's line of sight will be obstructed, the vehicle speed will decrease, the traffic flow will decrease accordingly, and the complexity of the traffic environment will increase; on snowy days, snowfall will cause snow to accumulate on the road, seriously affecting the road friction, making it difficult for vehicles to drive, the vehicle speed will drop significantly, the traffic flow will decrease sharply, and it may even cause traffic paralysis, and the traffic environment is extremely complex.

[0060] The traffic system division also includes intelligent driving guidance function space according to functional requirements Functional space for transportation support of major events The intelligent driving guidance function space mainly focuses on providing drivers with real-time and accurate driving route guidance information, and helps drivers avoid congested roads and improve travel efficiency by analyzing data such as traffic flow, vehicle speed, and road conditions; the major event traffic guarantee function space focuses on ensuring the orderly traffic around the event venue and related roads during major events, ensuring the smooth passage of vehicles and personnel related to the event, and minimizing the impact on surrounding daily traffic.

[0061] Step 1.2, constructing the virtual-real space traffic parameter vector, specifically includes the following two steps:

[0062] Step 1.2.1, real space traffic parameter construction: real space Ω R The traffic parameter vector at time t is in Traffic flow is collected through various traffic flow sensors installed on the road (such as geomagnetic sensors, radar sensors, etc.). Geomagnetic sensors use geomagnetic changes to sense the passing of vehicles. When a vehicle passes, the magnetic field changes, and the sensor records the number of changes to calculate traffic flow; radar sensors transmit electromagnetic waves and receive reflected waves, and calculate the speed and distance of the vehicle based on the time difference and frequency change of the reflected waves, thereby counting traffic flow. The vehicle speed can be obtained by a camera or an on-board sensor. The camera uses image recognition technology to analyze the position changes of the vehicle in continuous images to calculate the vehicle speed; the on-board sensor (such as a speed sensor) directly measures the vehicle's own driving speed. To increase the complexity of the traffic environment, the impact of weather (such as bad weather reducing road capacity) and road conditions (such as road construction, road damage, etc.) on traffic is comprehensively considered. Meteorological sensors monitor weather conditions (such as temperature, humidity, rainfall, wind speed, etc.). When bad weather occurs, such as heavy rain, heavy snow, fog, etc., it will reduce road friction, visibility, etc., thereby increasing the complexity of the traffic environment; road condition monitoring sensors (such as pavement sensors) detect whether there are potholes, construction, etc. on the road. If there is road construction, it will occupy the lane, reduce the road capacity, and thus increase the complexity of the traffic environment. These data are obtained through corresponding sensors and data processing to ensure the accurate acquisition of real-space traffic parameters.

[0063] Step 1.2.2, virtual space traffic parameter construction: virtual space The traffic parameter vector at time t is These parameters generate simulation values ​​based on preset models and algorithms. Traffic flow simulation can be calculated based on historical traffic data and traffic flow prediction models. For example, future traffic flow can be predicted based on historical data of the same period, recent traffic trends, and related traffic flow theoretical models (such as fluid dynamics models, following models, etc.); vehicle speed simulation considers factors such as road conditions (such as road slope, curvature, road surface conditions, etc.) and vehicle driving characteristics (such as vehicle type, power performance, etc.), and is generated by combining vehicle dynamics models and traffic flow models; traffic environment complexity simulation is determined by comprehensive weather simulation (such as simulating visibility and road friction under different weather conditions) and road condition simulation (such as simulating the impact of road construction, accidents, etc. on traffic). At the same time, relevant data of real space (such as real-time weather data) are introduced to calibrate it to improve the authenticity of the simulation and make the simulation of virtual space closer to the actual traffic situation.

[0064] Furthermore, in one embodiment, step 2 of constructing an interactive model of multiple parallel traffic spaces specifically includes:

[0065] Define from space To space The conversion function in It is the control quantity in real space (such as traffic light control parameters, speed limit instructions, etc.). Traffic light control parameters include green light duration, red light duration, yellow light duration, etc. Speed ​​limit instructions clearly stipulate the maximum speed of vehicles on the road; It is virtual space The simulated control quantity is generated according to the virtual space model and the control strategy. For example, when simulating traffic flow control, the simulated signal light control strategy and speed limit value are generated according to the preset optimization algorithm. The definition of the space conversion function is as follows:

[0066]

[0067] Among them, ω1, ω2, ω3 and ω4 are weight coefficients used to adjust the importance of each parameter in the conversion decision. The weight coefficients can be set according to different traffic scenarios and system requirements. For example, in a traffic congestion scenario, the weights of ω1 and ω2 (related to traffic flow and vehicle speed) may increase, because traffic flow and vehicle speed have a more critical impact on spatial conversion in congestion; in a severe weather scenario, the weight of ω3 (related to the complexity of the traffic environment) may increase to pay more attention to the triggering effect of weather factors on spatial conversion.

[0068] The interaction methods between different parallel spaces include:

[0069] Step 2.1, data exchange: data exchange between different parallel spaces according to demand, such as the intelligent driving guidance space exchanging traffic flow and speed data with the traffic congestion space to optimize the guidance route. The exchanged data must be encrypted and verified to ensure data accuracy and security. The frequency and content of data exchange can be dynamically adjusted according to traffic conditions;

[0070] Step 2.2, space transformation: When the transformation function satisfies When the space is converted, the conversion process involves the adjustment of the perception layer, data processing layer, control execution layer and model management layer. For example, when converting from the intelligent driving guidance space to the major event traffic guarantee space, the perception layer adjusts the sensor collection focus, the data processing layer re-determines the evaluation index and fusion weight, the control execution layer changes the control strategy, and the model management layer updates the model parameters;

[0071] Here, the conversion threshold θ ijPre-set according to different space conversion scenarios and traffic management needs. For example, in the conversion from sunny space to rainy space, thresholds can be set according to changes in key parameters such as rainfall and visibility. When these parameters reach or exceed the threshold, the space conversion is triggered. For example, when the weather changes from sunny to rainy, the speed and traffic flow in the real space change, and the sunny space The deviation between the simulated value and the real space observation value increases. If the corresponding conversion conditions are met, Convert to When a major event is about to be held, if you are currently in the intelligent driving guidance space If the traffic flow, vehicle speed and other parameters as well as control requirements meet the conversion conditions, it can be converted to the traffic support space for major events.

[0072] Step 2.3, collaborative iteration: Multiple parallel spaces work together to perform iterative optimization. For example, in traffic congestion management, the traffic congestion space, intelligent driving induction space, and major event traffic guarantee space (if there is an event) can collaboratively analyze the causes of congestion, jointly optimize traffic control strategies and induction information, and improve the system's ability to respond to traffic congestion through multiple iterations.

[0073] Furthermore, in one embodiment, the state equations and observation equations of modeling multiple parallel virtual and real spaces in step 3 specifically include:

[0074] Step 3.1, establish the state equation of virtual space at time t:

[0075] in, is the virtual space state vector, It is a virtual space simulation credibility index, used to measure the reliability of virtual space simulation results; It is a state transfer function built based on preset models (such as differential equations, dynamic system theory, and traffic flow models). The current state and observational data (including some observation data from real space and historical simulation data of virtual space itself, etc.) mapped to the state of the next moment; is the process noise.

[0076] The state transfer function It is constructed based on the understanding of the dynamic characteristics of the traffic system, and it is necessary to consider the change patterns of traffic parameters over time and the interactions between various factors in the traffic system. In the construction process, equations are used to describe the rate of change of traffic parameters, and the traffic system is regarded as a dynamic system. The evolution of the system is characterized by the changes in state variables. At the same time, combined with the vehicle following model and fluid mechanics model in traffic flow theory, the microscopic interactions between vehicles and the macroscopic behavior of traffic flow are simulated, so as to accurately reflect the transfer process of traffic status in virtual space. The state transfer function described in the present invention is defined as follows:

[0077]

[0078] The traffic parameter vector The state transfer part The expression is:

[0079]

[0080] in, Indicates the natural attenuation or growth trend of traffic flow, α1 (0<α1<1) is the attenuation coefficient. If α1 = 0.1, it means that when other factors remain unchanged, the traffic flow has a natural change (increase or decay) of 10% in each time step; It reflects the influence of vehicle speed on traffic flow. α2 is the speed influence coefficient. When the vehicle speed increases, the traffic flow may increase under certain conditions. For example, if α2 = 0.05, when the vehicle speed increases by a certain amount, the traffic flow will increase by a certain proportion. The specific increase depends on the change in vehicle speed and the value of α2. It reflects the inhibitory effect of traffic environment complexity on traffic flow. α3 is the influence coefficient of environmental complexity. For example, if α3 = 0.02, when the complexity of the traffic environment increases (such as bad weather), the traffic flow will decrease accordingly. It represents the direct influence of the control amount on the traffic flow, α4 is the influence coefficient of the control amount, if α4 = 0.1, the change of the control amount will cause the traffic flow to change proportionally; Considering the correction effect of observation data related to traffic flow on its prediction, α5 is the influence coefficient of observation data, and the observation data can come from partial data in real space or historical data in virtual space, etc.; It is process noise, reflecting the uncertainty in traffic flow forecasting, such as emergencies or model errors.

[0081] in, Indicates the natural change trend of vehicle speed, β1 (0<β1<1) is the natural change coefficient of vehicle speed. If β1=0.05, it means that the vehicle speed has a natural change of 5% in each time step; It reflects the impact of traffic flow on vehicle speed. β2 is the coefficient of traffic flow affecting vehicle speed. When the traffic flow is large, the vehicle speed may decrease. For example, β2 = -0.1. When the traffic flow increases, the vehicle speed will decrease proportionally. It reflects the limiting effect of the complexity of the traffic environment on the vehicle speed. β3 is the coefficient of environmental impact on vehicle speed. For example, if β3 = 0.03, the bad traffic environment will reduce the vehicle speed. It indicates the regulating effect of the control amount on the vehicle speed, β4 is the coefficient of the control amount affecting the vehicle speed. If β4 = 0.08, the change of the control amount will affect the vehicle speed; Considering the correction effect of the observation data related to vehicle speed, β5 is the influence coefficient of the vehicle speed observation data; It is process noise, reflecting the uncertainty in vehicle speed prediction.

[0082] in, Indicates the natural variation trend of the complexity of the traffic environment, γ1 (0<γ1<1) is the natural variation coefficient of the complexity of the environment, if γ1 = 0.05, the complexity of the traffic environment has a natural variation of 5% at each time step; and It reflects the impact of traffic flow and vehicle speed on the complexity of the traffic environment. γ2 is the coefficient of traffic flow affecting the complexity of the environment, and γ3 is the coefficient of vehicle speed affecting the complexity of the environment. For example, if γ2=0.02, an increase in traffic flow may increase the complexity of the traffic environment. If γ3=0.01, changes in vehicle speed will also affect the complexity of the traffic environment. It indicates the effect of control amount on the complexity of traffic environment. γ4 is the coefficient of control amount affecting the complexity of environment. If γ4=0.05, the change of control amount will affect the complexity of traffic environment. Considering the influence of observation data related to the complexity of the traffic environment, γ5 is the influence coefficient of observation data of environmental complexity; It is process noise, reflecting the uncertainty in the prediction of traffic environment complexity.

[0083] The control amount The state transfer part The expression is:

[0084]

[0085] in, Indicates the natural variation trend of the simulated control quantity, δ1 (0<δ1<1) is the natural variation coefficient of the control quantity. For example, if δ1=0.03, it means that the simulated control quantity has a natural variation of 3% in each time step; and Respectively reflect the impact of traffic flow, vehicle speed and traffic environment complexity on the simulation control quantity. δ2, δ3 and δ4 are the corresponding influence coefficients. For example, if δ2 = 0.05, when the traffic flow increases, the simulation control quantity will be adjusted according to a certain proportion. is the process noise corresponding to the simulated control quantity, which obeys the normal distribution with a mean of 0 (its covariance matrix is It is used to describe the uncertainty degree and correlation of process noise in the dimension of analog control quantity, determined according to system characteristics and data statistics), reflecting the uncertainty in the change of analog control quantity.

[0086] The virtual space simulation credibility index The state transfer part The expression is:

[0087]

[0088] in Indicates the natural variation trend of the simulated credibility index, ∈1 (0<∈1<1) is the natural variation coefficient of the credibility index. If ∈1=0.02, the simulated credibility index has a natural variation of 2% at each time step; and The simulation credibility index is adjusted according to the difference between the observed data and the simulated values ​​of traffic flow, vehicle speed and traffic environment complexity, ∈2, ∈3, ∈4 are the corresponding influence coefficients, N is the number of road sections, and M is the number of vehicles. For example, if ∈2 = 0.03, the greater the difference between the simulated value and the observed value of traffic flow, the greater the impact on the simulation credibility index; To simulate the process noise corresponding to the credibility index, it obeys a normal distribution with a mean of 0 (its covariance matrix is Determined based on system uncertainty and data statistics), reflecting the uncertainty in the prediction of simulation credibility indicators.

[0089] The coefficients α1-α5, β1-β5, γ1-γ5, δ1-δ4, ∈1-∈4, etc. in the above expressions need to be determined through historical traffic data, experimental data, and in-depth understanding of the traffic system. For example, α2 (the coefficient of vehicle speed affecting traffic flow) can be estimated by analyzing the change data of traffic flow at different vehicle speeds. When traffic is congested, the speed is reduced and the traffic flow will also decrease accordingly. Based on this relationship, the appropriate value of α2 can be determined. These coefficients reflect the mutual relationship and influence degree between various factors in the traffic system. Different traffic scenarios (such as congested sections in the city center, unobstructed sections on highways, etc.) may require different coefficient values ​​to accurately describe the state transition process. At the same time, the element values ​​of the covariance matrix of process noise reflect the source and degree of uncertainty in the traffic system, such as traffic flow measurement errors, randomness of traffic environment changes, etc. Reasonable estimates can be obtained through statistical analysis of a large amount of historical data, so that the virtual space state transfer function can more accurately simulate the dynamic changes of the traffic system in the virtual space.

[0090] The process noise in the above expression is and They are all assumed to obey a normal distribution with a mean of 0, and their covariance matrices are and It is used to describe the uncertainty degree and correlation of process noise in different state variable dimensions. The specific covariance matrix element values ​​are determined according to the characteristics of the traffic system and data statistical analysis.

[0091] Step 3.2: Establish the observation equation at time t in virtual space

[0092] in, is the observation vector, The components correspond to the observed values ​​of traffic flow, vehicle speed, and traffic environment complexity; is the observation function, which transforms the virtual space state vector Mapped to the observation space. Observation function Determined according to the simulation mechanism and observation method of the virtual space, for example, for traffic flow observation If the model-based calculation method is adopted, The simulated traffic flow values ​​are mapped to the observation space, which may include certain error correction terms to take into account the deviation between the model and the actual situation; if the observation is simulated by a virtual sensor, the mapping relationship is determined according to the simulation principle and performance parameters of the sensor; is the observation noise. The virtual space observation equation of the present invention is defined as follows:

[0093]

[0094] in, Represents the traffic flow in the virtual space state vector Speed and traffic environment complexity The main part of the calculated traffic flow observation value. α6 (0<α6<1) is the flow adjustment coefficient. For example, if α6=0.1, it means that the actually observed traffic flow may be 90% of the calculated value plus the influence of other factors. α7 is the influence coefficient of vehicle speed on flow observation, and α8 is the influence coefficient of traffic environment complexity on flow observation. They reflect the correction effect of vehicle speed and environmental complexity on traffic flow observation values. For example, a faster speed may cause the observed traffic flow to increase relative to the calculated value (α7 is a positive value), and a complex traffic environment may reduce the observed flow (α8 is a negative value). is the fixed part of the observation error, which may be caused by the systematic deviation of the virtual space model from the traffic flow observation, such as the fixed deviation caused by the difference between the model assumptions and the actual traffic behavior. If it is found that the traffic flow simulated in the virtual space is always 5 vehicles / minute larger than the actual observation value (assuming the traffic flow unit), then is the observation noise, assuming that it has a mean of 0 and a covariance matrix of The normal distribution of is used to represent the random errors in traffic flow observations, such as random fluctuations in sensor measurements, random interference in data transmission, etc. The covariance matrix The element values ​​of are determined according to factors such as sensor accuracy and the stability of the observation environment. For example, when the sensor accuracy is high, the element values ​​of the covariance matrix are relatively small, indicating that the observation noise is small.

[0095] in, The main part of the speed observation value is calculated based on the virtual space state. β6 (0<β6<1) is the speed adjustment coefficient. If β6=0.05, it means that the actual observed speed is 95% of the calculated value plus other influences. β7 is the influence coefficient of traffic flow on speed observation, and β8 is the influence coefficient of traffic environment complexity on speed observation. For example, heavy traffic flow may reduce the observed speed (β7 is a negative value), and complex traffic environment (such as slippery roads) will reduce the speed observation value (β8 is a negative value). is the fixed part of the speed observation error, which may be caused by the model's inaccurate assumptions about the speed observation. For example, if the model always overestimates the speed by 2 km / h, then is the observation noise, with mean 0 and covariance matrix The normal distribution of reflects the random error in vehicle speed observation, and the value of its covariance matrix elements depends on factors such as the accuracy of the vehicle speed sensor and the measurement environment.

[0096] in, The main part of the traffic environment complexity observation value is calculated according to the virtual space state. γ6 (0<γ6<1) is the environmental complexity adjustment coefficient. If γ6=0.03, it means that the actual observed environmental complexity is 97% of the calculated value plus the influence of other factors. γ7 is the influence coefficient of traffic flow on environmental complexity observation, and γ8 is the influence coefficient of vehicle speed on environmental complexity observation. For example, heavy traffic flow or slow vehicle speed may increase the traffic environment complexity observation value (γ7 and γ8 are positive values). is the fixed part of the observation error of traffic environment complexity, which may be caused by the inaccurate assessment of environment complexity by the model. For example, if the model's assessment of the impact of road construction is always too high, resulting in the observed value of environment complexity always being 0.1 greater than the actual value (assuming that environment complexity is a quantitative indicator), then is the observation noise, with mean 0 and covariance matrix The normal distribution of reflects the random error in the observation of traffic environment complexity. The element values ​​of its covariance matrix are determined according to factors such as the accuracy of environmental monitoring sensors and the uncertainty of environmental changes.

[0097] The coefficients α6-α8, β6-β8, γ6-γ8, etc. in the above expression need to be determined according to the characteristics of the virtual space model, the characteristics of the observed data, and the behavior of the actual traffic system. These coefficients can be estimated by comparing and analyzing a large amount of historical observation data and virtual space simulation data. For example, the values ​​of α6-α8 can be determined by analyzing the relationship between the actual observed traffic flow and the virtual space calculated value under different traffic flows and vehicle speeds. The fixed part of the observation error It needs to be determined based on the analysis of model deviations and the statistics of actual observation data, and the covariance matrix of the observation noise is It is determined based on factors such as sensor accuracy and observation environment stability. These parameters work together to enable the virtual space observation function to reflect the actual observation situation as accurately as possible on the basis of considering the virtual space state, while taking into account various errors and uncertainties in the observation process, providing a reliable observation data basis for the analysis and management of the transportation system.

[0098] Step 3.3, establish the observation equation at time t in real space:

[0099] in is the real space observation vector, is the real-space observation function, which is determined according to the measurement principles and data processing methods of various sensors in the real space, and converts the real-space traffic parameter vector Mapped to the observation space. For example, for traffic flow observation If measured with a radar sensor, The actual traffic flow is converted into an observed value based on the physical relationship between the radar signal and the traffic flow, which may involve signal processing and calibration algorithms to improve the measurement accuracy. If a geomagnetic sensor is used, the mapping method is determined based on the relationship between geomagnetic changes and vehicle passage. It is the real space observation noise. This way of establishing the observation equation can accurately reflect the actual state of the real space through sensor data. Although the internal state equation of the real space is unknown, sufficient information can be obtained through the observation equation for subsequent system analysis and control. Since the operation mechanism of the real space is difficult to obtain, the real space state equation is unknown, but the real space can be analyzed by observation. The real space observation equation described in the present invention is defined as follows:

[0100]

[0101] in, is the traffic flow observation scale factor, and its value depends on the type and characteristics of the traffic flow sensor. For example, if a geomagnetic sensor is used, The value of will be determined based on the sensitivity of the geomagnetic sensor to the passing of vehicles and the data conversion algorithm. Assuming that the geomagnetic sensor is calibrated, each time a certain intensity of geomagnetic change is sensed, a certain number of vehicles pass through. Based on this relationship, If radar sensors are used to measure traffic flow, It is determined based on the relationship between the radar signal reflection principle and traffic flow, that is, it is calculated based on the relationship between the time difference between the radar transmitting electromagnetic waves and receiving reflected waves, frequency changes, and the number and speed of vehicles. The value of . It is the fixed part of the traffic flow observation error, which may be caused by factors such as the sensor installation location and environmental interference. For example, if there is a large metal object interference near the geomagnetic sensor, the measured value may always be larger or smaller than the actual traffic flow by a fixed value. If it is found that the measured value is always larger than the actual value by 3 vehicles / minute (assuming the traffic flow unit), then is the traffic flow observation noise, assuming that its mean is 0 and the covariance matrix is The normal distribution of is used to represent the random errors in traffic flow observations. The covariance matrix The element values ​​of are determined according to factors such as sensor accuracy, random fluctuation characteristics of traffic flow, and uncertainty of environmental interference. For example, when the traffic flow is large and the vehicle speed is uneven, the random error of traffic flow observation may be large, which is reflected in the covariance matrix The value of the middle element is larger.

[0102] in, is the vehicle speed observation ratio coefficient, which is determined by the vehicle speed sensor. If a camera is used for vehicle speed observation, The speed of a vehicle is determined by the relationship between the vehicle position change and time in image recognition technology and the camera frame rate. For example, the camera calculates the vehicle speed by analyzing the displacement of the vehicle in continuous images and the shooting time interval. It is to convert the displacement and time relationship in the image into the proportional coefficient of the actual vehicle speed. If the vehicle-mounted sensor (such as a speed sensor) is used to measure the vehicle speed, It is determined according to the sensor's measurement principle and data conversion relationship. Usually, the vehicle-mounted speed sensor directly measures physical quantities such as wheel speed, and then converts it into vehicle speed through parameters such as vehicle tire circumference. This is the proportional coefficient in this conversion process. This is the fixed part of the speed observation error, which may be caused by the sensor calibration error or the influence of environmental factors on the sensor measurement. For example, if the camera is not in a good shooting angle or the lighting conditions are unstable, it may cause a fixed deviation in the measured speed. If the speed measurement is always 2km / h lower than the actual value, then is the speed observation noise, which has a mean of 0 and a covariance matrix of The normal distribution of reflects the random error in the speed observation. The element values ​​of its covariance matrix depend on the accuracy of the speed sensor, the speed fluctuation during the vehicle driving process (such as the randomness of the driver's acceleration and deceleration behavior), and environmental factors (such as the speed measurement error caused by the change of road slope). For example, on a road section with a large change in road slope, the random error of the speed observation may increase, which is reflected in the covariance matrix The value of an element changes.

[0103] in, is the observation ratio coefficient of traffic environment complexity, which is determined according to the definition and monitoring method of traffic environment complexity. For example, if the traffic environment complexity mainly considers weather conditions and road construction conditions, for weather conditions, the temperature, humidity, rainfall, wind speed and other data measured by meteorological sensors are determined through a certain conversion relationship If the rainfall reaches a certain level, its weight in the complexity of the traffic environment is determined according to the degree of impact on traffic, thus determining For road construction, when road condition monitoring sensors (such as pavement sensors) detect road construction, they determine the degree of impact on the complexity of the traffic environment based on factors such as construction type and construction scope, and then determine is the fixed part of the traffic environment complexity observation error, which may be caused by the inaccurate sensitivity of the sensor to environmental factors or the deviation in the data processing process. For example, if the meteorological sensor has a deviation in the measurement of rainfall, it may cause a fixed error in the traffic environment complexity observation value. If it is found that due to sensor problems, the traffic environment complexity observation value is always 0.05 lower than the actual situation in rainfall (assuming that traffic environment complexity is a quantitative indicator), then is the traffic environment complexity observation noise, which has a mean of 0 and a covariance matrix of The normal distribution of reflects the random errors in the observation of traffic environment complexity. The values ​​of its covariance matrix elements are determined according to factors such as the accuracy of environmental monitoring sensors, the uncertainty of weather changes (such as the difficulty in accurately measuring sudden changes in the intensity of rainstorms), and the unpredictability of temporary events such as road construction (such as the uncertainty of the impact of changes in construction progress on traffic).

[0104] The traffic flow observation proportionality coefficient in the above expression Vehicle speed observation ratio and traffic environment complexity observation ratio coefficient It needs to be determined based on the specific type of sensor, installation method, calibration and the actual physical relationship of the traffic system. The accurate values ​​of these coefficients can be obtained through precise calibration of the sensor, experimental testing and analysis of a large amount of actual observation data. The fixed part of the observation error It is necessary to conduct a detailed analysis of the sensor's working environment, installation conditions, and historical observation data to identify and quantify the factors that may cause fixed deviations. It is determined by a comprehensive evaluation of factors such as sensor accuracy, random characteristics of the traffic system, and environmental interference. These parameters work together to enable the real-space observation function to accurately convert the actual traffic state in the real space into observation values, while taking into account various errors and uncertainties in the observation process, providing reliable data support for the management and decision-making of the traffic system. For example, an accurate observation function helps traffic management departments to reasonably adjust the timing of traffic lights based on actual traffic flow and vehicle speed observations, and to timely issue traffic warning information based on traffic environment complexity observations.

[0105] Furthermore, in one of the embodiments, the specific implementation steps of step 4 are as follows: through the five-stage steps of perception, evaluation, diagnosis, optimization and implementation, closed-loop management of the transportation system is achieved, and each stage works closely together to continuously optimize system performance based on multi-parallel space models and virtual and real space parameter solutions to adapt to changes in the traffic environment.

[0106] The specific steps include:

[0107] Step 4.1, perception phase: collect and fuse real space and virtual space data to provide a data basis for subsequent operations, while monitoring information related to the interaction between different parallel spaces;

[0108] Step 4.2, evaluation phase: Calculate multiple evaluation indicators and conduct spatial evaluation to determine the virtual space simulation effect and evaluate the impact of different parallel space interactions on the transportation system;

[0109] Step 4.3, diagnosis phase: trigger diagnosis based on specific thresholds to analyze the source of virtual space or real space problems;

[0110] Step 4.4, optimization phase: optimize virtual space parameters and real space control strategy according to the diagnosis results;

[0111] Step 4.5, implementation phase: Apply optimization strategies, update the real and virtual space states and observation vectors, prepare for the next round of closed loop, and ensure the effective execution of interactions between different parallel spaces.

[0112] Preferably, in some embodiments, the specific implementation steps of step 4.1 are as follows:

[0113] Step 4.1.1, Data Collection: In real space Ω R In the process, traffic parameter data is collected through various sensors (such as geomagnetic sensors, radar sensors, cameras, vehicle-mounted sensors, meteorological sensors, etc.) to obtain real-space observation vectors. The geomagnetic sensor detects the change of geomagnetic field to sense the passing of vehicles and obtain traffic flow. Related data; radar sensors use the principle of electromagnetic wave reflection to measure vehicle distance and speed, and further calculate traffic flow and vehicle speed The camera uses image recognition technology to analyze the vehicle's position and motion status to obtain vehicle speed and traffic flow information; on-board sensors (such as speed sensors, acceleration sensors, etc.) directly obtain the vehicle's own driving status information, such as speed, acceleration, etc. Acceleration information can be used to analyze the vehicle's acceleration and deceleration behavior, and then judge the smoothness and congestion of traffic; meteorological sensors monitor weather conditions (such as temperature, humidity, rainfall, wind speed, etc.) to analyze the impact of weather on the complexity of the traffic environment These sensors acquire various information of the traffic system in real time according to their respective working principles and installation locations, ensuring the comprehensiveness and accuracy of the data. In the process, simulated traffic parameter data is generated according to the preset model and algorithm, and combined with some real space observation data (such as weather data, etc.), the virtual space state vector is obtained. and the observation vector The data generation of the virtual space is based on its internal mathematical model and the simulation logic of different traffic scenarios. For example, traffic flow simulation may be based on historical traffic data and traffic flow prediction models, vehicle speed simulation takes into account vehicle dynamics models and road condition simulations, and traffic environment complexity simulation is determined by comprehensive weather simulation and road condition simulation. At the same time, relevant data of the real space (such as real-time weather data) is introduced to improve the authenticity of the simulation. In addition, data related to the interaction between different parallel spaces, such as space conversion requests, data exchange records, etc., are collected for subsequent analysis of the interaction between different parallel spaces.

[0114] Step 4.1.2, data fusion: Fusion of traffic parameters. Taking traffic flow fusion as an example, let the fused traffic flow observation value be The calculation formula is: where α Q,t and It is a weight that is dynamically adjusted according to the sensor accuracy, virtual space simulation credibility and current traffic environment, and satisfies The accuracy of the sensor is determined by the calibration and performance evaluation of the sensor. The weight of the sensor with high accuracy is relatively large during fusion. For example, a radar sensor that has been accurately calibrated and has a small error has a high accuracy in measuring traffic flow, and its weight will increase accordingly. The credibility of the virtual space simulation is evaluated based on the accuracy of the virtual space model and the historical simulation effect. When the credibility is high, the weight of the corresponding data increases. If the virtual space model fits the real space data well in past simulations and predicts the traffic conditions more accurately, then the weight of its simulation data will increase during fusion. The current dynamic adjustment of the traffic environment takes into account factors such as the degree of traffic congestion and weather changes. For example, during congestion, it may rely more on real space observation data, and the weight is adjusted accordingly. When traffic congestion is severe, the real-time traffic flow data directly obtained by the real space sensor can better reflect the actual situation. At this time, the weight of the real space observation data α Q,t will increase, and the weight of the virtual space simulation data will decrease. Similarly, the corresponding fusion operation is performed on the data such as vehicle speed and traffic environment complexity to obtain the fused observation vector This comprehensive data fusion provides a rich and accurate data foundation for subsequent evaluation, diagnosis and optimization operations.

[0115] Preferably, in some embodiments, the specific implementation steps of step 4.2 are as follows:

[0116] Step 4.2.1, Evaluation Index Calculation: Calculate the Traffic Congestion Index Average travel time T t , virtual and real space fitting degree index MAE t , Accuracy of data exchange between different parallel spaces data-exchangeand space conversion success space-transfer Indicators such as , are used to evaluate the effectiveness and stability of interactions between different parallel spaces. The specific definitions of each indicator are as follows:

[0117]

[0118] In formula (8), N is the number of road sections. and They are the maximum traffic flow, vehicle speed and traffic environment complexity of road section j under ideal conditions, which can be determined based on historical data or preset standards. The traffic congestion index can intuitively reflect the current traffic congestion situation and provide an important basis for system evaluation. For example, when the traffic flow approaches or exceeds the maximum traffic flow of the road section, the vehicle speed is significantly lower than the ideal speed, and the traffic environment is more complex (such as bad weather or road construction), the traffic congestion index will increase.

[0119]

[0120] In formula (9), M is the number of vehicles, and the vehicle entry and exit times are and (Acquired through vehicle positioning data). Average travel time is one of the key indicators for measuring traffic efficiency. By calculating and analyzing it, we can understand the changes in the operating efficiency of the traffic system. If the average travel time increases, it means that the operating efficiency of the traffic system has decreased, and there may be congestion or other traffic problems.

[0121]

[0122] The index defined in formula (10) is used to measure the degree of fit between the virtual space simulation results and the actual observation results in the real space, and is an important basis for evaluating the performance of the virtual space. t The smaller the value, the closer the virtual space simulation is to the actual situation of real space, and the higher the accuracy of the virtual space model.

[0123]

[0124] In formula (11), Number correct-exchange Number is the number of correctly exchanged data. total-exchange is the total number of data exchanges. A high data exchange accuracy rate indicates that the data exchange between parallel spaces is reliable and can provide accurate information for the system.

[0125]

[0126] In formula (12), Number success-transfer Number is the number of spaces successfully converted. request-transferis the number of space conversion requests. A high space conversion success rate indicates that the system operates effectively in different space conversions and can adapt to different traffic scene requirements.

[0127] Step 4.2.2, spatial evaluation based on evaluation indicators: Based on the calculated evaluation indicators, compare the differences in evaluation indicators between different virtual spaces and real spaces, such as calculating the difference in traffic congestion index between each virtual space and real space. Average travel time difference Differences in virtual and real space fitting degree index Through these differences, the simulation effect of virtual space on real space is evaluated. If the difference is within a certain range and the trend is consistent with the trend of traffic conditions in real space, it means that the simulation effect of virtual space is good; if the difference is too large or the trend is inconsistent, further analysis and adjustment may be required. For example, if the traffic congestion index of virtual space is similar to that of real space, and changes accordingly with the changes in traffic congestion conditions in real space, it means that the simulation of congestion conditions in virtual space is more accurate. At the same time, the impact of the interaction of different parallel spaces on the overall traffic system is evaluated. If the interaction-related indicators are abnormal, analyze their impact on the trend of traffic conditions. For example, if the data exchange accuracy is low, the real space data obtained by the virtual space may be inaccurate, thereby affecting the simulation results of the virtual space, and then affecting the adjustment of the traffic control strategy based on this, which may eventually lead to increased traffic congestion or reduced traffic efficiency. If the space conversion success rate is low, problems such as data loss or system instability may occur during the conversion process, affecting the normal operation of the traffic system.

[0128] Preferably, in some embodiments, the specific implementation steps of step 4.3 are as follows:

[0129] Step 4.3.1, Diagnosis trigger mechanism setting: When MAE t >θ t (where θ t is a threshold value dynamically adjusted according to different traffic scenarios and system performance requirements), the diagnosis process is triggered. t The dynamic adjustment method is as follows: In different traffic scenarios, such as normal traffic flow, traffic congestion, bad weather, special events, etc., the virtual-real space fitting degree index MAE in different scenarios is calculated based on historical data t In the normal traffic flow scenario, the MAE in the past p time periods (p can be set according to the actual situation, such as p = 20) t The mean value of The standard deviation is σ normal The threshold θ can be set t for (k is a coefficient set according to the accuracy requirements of the system, such as k = 2). In normal traffic flow, the traffic conditions are relatively stable and vehicles drive regularly, so MAE tThe fluctuation of MAE is small, and the threshold is set relatively low to ensure that potential problems are discovered in time. t It may increase. In this case, different thresholds can be set according to the degree of congestion. For example, in the case of mild congestion, the threshold is calculated based on historical mild congestion data. and σ mild-congestion , the threshold is set to (k mild The tolerance for accuracy can be set according to the light congestion); the same applies to heavy congestion. In congestion, traffic parameters vary greatly, and vehicle driving is affected by many factors. MAE t The range of change will also increase, so it is necessary to set appropriate thresholds according to different congestion levels to avoid misjudgment or missed judgment. For special scenarios such as severe weather, appropriate thresholds are also determined based on corresponding historical data. Severe weather such as heavy rain, heavy snow, and heavy fog will have a significant impact on traffic, and parameters such as vehicle speed and traffic flow will change significantly. It is necessary to set appropriate thresholds based on historical data of MAE under severe weather conditions. t The distribution of the data can be used to set a threshold to accurately determine whether there is a problem with the system.

[0130] Step 4.3.2, Problem source analysis: Calculate the relative probability P of the virtual space model being inaccurate r The relative probability P of interference between (H1) and real space r (H2) = 1-P r (H1), according to P r (H1) and P r (H2) can be used to initially determine the source of the problem. If P r (H1) is large, the virtual space model may be inaccurate; if P r If (H2) is large, there may be interference in the real space. If the virtual space model is inaccurate, further analysis of the virtual space state transfer function and the observation function Parameters of traffic flow related state transition function Check parameter α 1 , α 2 , α 3 , α 4 , α 5 For example, if it is found that the traffic flow simulated in the virtual space deviates greatly from the observed value in the real space under traffic congestion, α 2 Whether the (speed-related parameters) correctly reflect the impact of vehicle speed on traffic flow during congestion, it may be necessary to refit the parameters based on actual congestion data. Checking for observation errors If the traffic flow observation value is always larger or smaller than the actual value by a fixed value, the estimation can be adjusted. . Similarly, the parameters of the state transfer function and observation function related to vehicle speed, traffic environment complexity, etc. are analyzed and adjusted. If there is interference in the real space, analyze the interference source. For example, in the bad weather space, check whether the meteorological sensor is working properly, whether other sensors (such as geomagnetic sensors measuring traffic flow affected by waterlogging, etc.) are inaccurate due to weather reasons, etc. For meteorological sensors, their measurement accuracy can be checked by comparing with the data of other nearby meteorological stations. If the rainfall, temperature and other data measured by the meteorological sensor are significantly different from those of the surrounding meteorological stations, there may be a problem. For the case where the geomagnetic sensor is affected by waterlogging, the measurement results can be analyzed and corrected based on the geomagnetic signal processing principle and the relationship model between the depth of waterlogging and the measurement error of traffic flow (established through experiments or historical data). At the same time, check whether there are other temporary interference factors on the road, such as road construction, traffic accidents, etc. that affect the measurement of traffic parameters. If there is road construction, the degree of its impact on traffic flow, vehicle speed and traffic environment complexity can be analyzed based on information such as the length of the construction section, the type of construction and the construction time. For example, when the construction section is long and the construction type is fully closed, the impact on traffic flow may be greater, resulting in a decrease in traffic flow; when the construction time is during the peak traffic period, the impact on traffic congestion is more obvious. r (H1) is defined as follows:

[0131]

[0132] in Z Q,ref is the reference traffic flow value (the value can be the traffic flow corresponding to the road design capacity or the average traffic flow in the same period according to history), Z V,ref is the reference speed value (the value can be the road speed limit, historical average speed, etc.), Z C,ref is the reference value of the complexity of the traffic environment (which can be determined comprehensively based on factors such as weather conditions and road construction conditions), and l is the scaling factor.

[0133] Step 4.3.3, further analysis and adjustment: If the virtual space model is inaccurate, in addition to the above parameter analysis, the model structure diagnosis is also required. Model structure identification technology (such as information criterion-based methods, Bayesian model selection, etc.) is used to determine whether the structure of the current virtual space model is reasonable. For example, if the traffic data presents a complex nonlinear relationship, but the virtual space model adopts a simple linear structure, the data fitting effect under different model structures is compared by information criteria (such as AIC, BIC, etc.) to determine whether the model structure needs to be adjusted. For the analysis of real space interference sources, if there is road construction, in addition to considering the impact of the length of the construction section, the type of construction, and the construction time on the surrounding traffic flow and vehicle speed, the occupation of construction equipment, the activity range of construction personnel, and other factors on the sensor measurement can be further analyzed, as well as the impact of dust and noise generated during the construction process on the traffic environment, so as to more comprehensively evaluate the real space interference and take corresponding measures. For example, the occupation of construction equipment may cause the sensor to misjudge the position and speed of vehicles or pedestrians, the dust generated by construction may affect the image recognition accuracy of the camera, and the noise may interfere with the normal operation of vehicle sensors or roadside sensors. If the camera image is blurred due to dust cover, it will affect its recognition of the vehicle and road conditions, and thus affect the collection of data such as vehicle speed and traffic flow; vehicle-mounted sensors may be interfered with by noise, resulting in inaccurate measurement data, affecting the judgment of the vehicle's driving status.

[0134] Preferably, in some embodiments, the specific implementation steps of step 4.4 are as follows:

[0135] The virtual space parameter optimization aims to improve the simulation accuracy of the virtual space model for actual traffic conditions. Specifically, the virtual and real space fitting degree index MAE is minimized. t Assume that the objective function is Here Represents virtual space These parameters directly affect the simulation effect of the virtual space model on the traffic system, including the simulation of traffic flow, vehicle speed, and traffic environment complexity.

[0136] To achieve this goal, a gradient-based optimization algorithm is used, such as the gradient descent algorithm, which calculates the gradient of the objective function with respect to the parameters and gradually adjusts the parameters according to the direction and magnitude of the gradient to find the parameter combination that minimizes the objective function.

[0137] Step 4.4.1, virtual space parameter optimization: Optimize virtual space parameters using gradient-based optimization algorithm To minimize the virtual-real space fitting degree index MAE t Assume that the objective function is According to the gradient descent algorithm, the parameter update formula Where α is the learning rate, is the objective function About parameters When calculating the gradient, it is necessary to use the chain rule to find out the influence of each parameter on the objective function based on the virtual space state equation and the observation equation. In the optimization process, an adaptive learning rate strategy is introduced. The learning rate α is adjusted according to the change of the objective function in each iteration. Assume if and If the product of is greater than 0, it means that the learning rate may be too large, and α can be appropriately reduced; if the product is less than 0, it means that it may be trapped in a local optimum, and you can try to increase α or use other optimization strategies (such as simulated annealing algorithm, etc.) to jump out of the local optimum. In addition, you can also combine swarm intelligence algorithms (such as ant colony algorithm, particle swarm optimization algorithm, etc.) with gradient-based optimization algorithms. First, use the swarm intelligence algorithm to perform a preliminary search on the virtual space parameters to obtain a better parameter range, and then use the gradient-based algorithm within this range for fine optimization to improve the optimization effect.

[0138] The chain rule mentioned above takes the optimization of traffic flow related parameters as an example. For the objective function Medium traffic flow i.e. MAE t The part related to traffic flow set up The parameters related to traffic flow are (corresponding to the coefficients α2, α3, etc. in the previous state transfer equation, in order to distinguish different virtual spaces and their relationship with other parameters, we use The specific optimization steps are as follows:

[0139] Step 4.4.1.1, first, according to the virtual space traffic flow observation equation (in is the fixed part of the observation error, is the observed noise) and the fused traffic flow observations Find From the observation equation we can get (Assumption and No matter, and irrelevant), then

[0140] Step 4.4.1.2, according to the virtual space traffic flow state transfer equation (Here the coefficients α2 and α3 are Expressed as , to reflect the optimization of parameters related to traffic flow), find For example, Find partial derivatives right Find partial derivatives right Find partial derivatives (Assumption It has nothing to do with other variables in the state transfer equation. This is only to fully demonstrate the partial derivative calculation process. If there is a correlation, the partial derivative needs to be calculated based on the actual relationship).

[0141] Step 4.4.1.3, then by the chain rule Find Then get Similarly, the gradient of the objective function of parameters such as vehicle speed and traffic environment complexity can be calculated to achieve the virtual space parameters (including traffic flow related Optimization of parameters such as .

[0142] Step 4.4.2, real space control strategy optimization: adjust the real space control strategy according to the virtual space optimization results For example, if the virtual space optimization indicates that the current traffic congestion is aggravated, the signal light control parameters in the real space can be adjusted accordingly to extend the green light time in the congested direction. The specific adjustment method is as follows: Assume that the signal light control parameters include the green light time G t 、Red light duration R t 、Yellow light duration t , total period t = G t +R t +Y t According to the traffic congestion direction d (the congestion direction can be determined by traffic flow and vehicle speed data), G t (d) The value of the increase ΔG t (d) = k G C congestion (d)(k G is a coefficient determined based on actual experience). At the same time, the speed limit instruction can be adjusted appropriately. Suppose the original speed limit value is V limit , according to the speed and traffic flow conditions simulated in the virtual space, the new speed limit value V new-limit =V limit -k V (C congestion ×V average )(k V is the coefficient, V average= is the current average vehicle speed). If it is in a traffic guarantee space for major events, the signal light timing can be adjusted according to factors such as the scale of the event, the number of participants, and the road conditions around the event venue, and special lanes or priority rights of way can be set up for vehicles related to the event, and temporary parking areas can be set up to guide vehicles to park in an orderly manner. For example, before and after the event, the green light duration of the roads around the event venue leading to the main traffic artery can be increased, and the green light duration in other directions can be reduced to guide vehicles to evacuate quickly; during the event, the signal light cycle can be dynamically adjusted according to the real-time traffic flow at the event site to ensure the orderly passage of vehicles related to the event and ordinary traffic vehicles. For intelligent driving guidance space, according to real-time traffic conditions and road information, the driver is provided with guidance information such as the optimal driving route. For example, the on-board navigation system or roadside variable message signs (VMS) can be used to recommend routes to drivers that avoid congested sections, provide information such as estimated travel time and real-time road conditions, guide drivers to reasonably choose driving routes, and improve road traffic efficiency.

[0143] Preferably, in some embodiments, the specific implementation steps of step 4.5 are as follows:

[0144] Step 4.5.1, control strategy implementation: Apply the optimized real-space control strategy Applied to real traffic systems, such as real-time control of traffic through traffic control equipment (signal lights, variable speed limit signs, etc.). For signal light control, the traffic control center sends new signal light control parameters (green light duration, red light duration, yellow light duration) to the signal light controllers at each intersection, and the signal light controllers adjust the on and off time of the signal lights according to the received instructions. In the process of sending instructions, a reliable communication protocol (such as the NTCIP protocol) is used to ensure the accuracy and stability of data transmission. At the same time, in order to deal with abnormal situations such as communication failures, the signal light controller can set a local backup control strategy. When it loses contact with the traffic control center, it operates according to the local backup strategy to avoid traffic chaos. For variable speed limit signs, new speed limit instructions are sent to the sign controller through wireless communication technologies (such as LTE-V, DSRC, etc.), and the sign controller updates the displayed speed limit value. In the process of wireless communication, encryption and error correction coding technologies are used to prevent instructions from being tampered with and transmission errors. For example, the AES encryption algorithm is used to encrypt the speed limit instruction, and the error correction coding method such as Hamming code is used to ensure the correctness of the data. If the traffic is guaranteed in a space for a major event, control strategies can be implemented through on-site command by traffic police and the installation of traffic control facilities (such as roadblocks, traffic cones, etc.) to ensure orderly traffic during the event.

[0145] Step 4.5.2, system status update: real space status update, according to the actual operation of the real space, re-collect data through the sensor and update the real space observation vector And according to the real space observation equation, the real space traffic parameter vector is inferred (Although the real-space state equation is unknown, the estimate of the real-space state can be continuously updated through observation data.) For example, the radar sensor continues to scan the road to obtain new traffic flow and vehicle speed information, and updates and Weather sensors update weather conditions information for updates Then according to the real space observation equation Reverse deduction through numerical calculation methods (such as least squares method, etc.) The virtual space state is updated according to the virtual space state equation and combined with the updated observation data. Update virtual space state vector And the new virtual space observation vector is obtained through the virtual space observation equation For example, according to the virtual space traffic flow state transition equation The updated speed Traffic environment complexity Control volume and new observational data Substitute into the equation and calculate the new traffic flow Similarly, update the parameters such as vehicle speed and traffic environment complexity to update the virtual space state vector According to the virtual space observation equation Get the new virtual space observation vector

[0146] Step 4.5.3, preparation for a new round of closed-loop operation: Based on the new real-space observation vector and the virtual space observation vector Re-perform closed-loop operations such as data fusion, evaluation, diagnosis, and optimization to enable the system to continuously adapt to the dynamic changes in the traffic environment and continuously optimize system performance. The specific implementation steps are as follows:

[0147] Step 4.5.3.1, data fusion: Recalculate the fused traffic flow observations using the same fusion method as before The formula is The weight α Q,t+1 and According to the new sensor accuracy, virtual space simulation credibility and the current traffic environment, dynamic adjustments are made. Corresponding fusion operations are also performed on data such as vehicle speed and traffic environment complexity to obtain a new fused observation vector At the same time, the new data related to the interaction between different parallel spaces are sorted and preprocessed.

[0148] Step 4.5.3.2, Evaluation phase: Recalculate the traffic congestion index Average travel time T t+1 And the virtual and real space fitting degree index MAE t+1 At the same time, new evaluation indicators of the interaction effects of different parallel spaces are calculated, such as the updated data exchange accuracy data-exchange,t+1 and space conversion success space-transfer,t+1 Then, space evaluation is performed based on these new evaluation indicators, and the differences in new evaluation indicators between different virtual spaces and real spaces are compared, such as and And analyze the new impact of different parallel space interactions on the overall transportation system to determine whether the virtual space simulation effect is still good.

[0149] Step 4.5.3.3, Diagnosis phase: Check whether the new diagnostic trigger condition, i.e. MAE, is met t+1 >θ t+1 (where θ t+1 is a threshold value that is dynamically adjusted according to new traffic scenarios and system performance requirements). If the diagnosis is triggered, the relative probability P of the virtual space model being inaccurate is recalculated. r The relative probability P of interference between (H1, t+1) and real space r (H2, t+1), analyze the source of the problem and determine whether it is a virtual space model problem or a real space interference problem. If the problem is that the virtual space model is inaccurate, further analyze the virtual space state transfer function and the observation function If there is interference in the real space, check for new interference sources, such as new road construction, temporary traffic control, or sensor failure.

[0150] Step 4.5.3.4, Optimization phase: Based on the new diagnostic results, if optimization is required, the gradient-based optimization algorithm is used again to optimize the virtual space parameters. To minimize the new virtual-real space fitting degree index MAE t+1 is the objective function According to the gradient descent algorithm Update the parameters, where the learning rate α t+1 According to the convergence of the optimization process, the gradient Calculate through the new virtual space state equation and observation equation. At the same time, adjust the real space control strategy according to the new optimization results of the virtual space Such as further adjusting traffic light control parameters, speed limit instructions, etc.

[0151] Step 4.5.3.5, Implementation phase: Apply the new optimized real-space control strategy Applied to real traffic systems, new traffic control is achieved through traffic control equipment. At the same time, the updated system state operation is performed, that is, according to the new observation vector in real space Update the real-space traffic parameter vector According to the new observation vector of virtual space Update virtual space state vector Prepare for the next round of closed-loop operations and continuously optimize the operation of the transportation system. During the implementation process, monitor the execution of the control strategy to ensure that the transportation equipment responds to instructions accurately and handle and adjust abnormal situations in a timely manner.

[0152] The virtual-real space conversion function T of the present invention ij , virtual space state transfer function Virtual space observation function Real space observation function Function expressions only provide a specific and feasible implementation method. Different forms of expressions can be used in different scenarios, which does not affect the main implementation logic of the parallel transportation system with virtual-reality interaction and closed-loop drive described in the present invention.

[0153] As a specific example, the present invention is further described in detail in some of the embodiments.

[0154] Example 1: Traffic congestion space (such as )Implementation Example

[0155] Embodiment 1 of the present invention is a parallel space for traffic congestion (such as ), the specific implementation steps are as follows:

[0156] Step (1) Perception stage: In real space Ω R In the process, sensors collect traffic flow Increase, speed Reduced data and traffic environment complexity The number of vehicles has increased due to the increase in traffic and road congestion. On-board sensors show that the frequency of sudden acceleration and braking of vehicles has increased (which can be determined by accelerometer data), as well as the length of vehicle queues (which can be estimated by combining radar sensor and camera data). In the virtual space, the prediction model is adjusted according to the congestion situation, taking into account factors such as vehicle queues and delays, and generating new traffic flows. Speed Traffic environment complexity The simulated value of the virtual space state vector and the observation vector In the data fusion stage, the weights are adjusted according to the sensor accuracy and the credibility of the virtual space for congestion simulation. Since congestion may affect the accuracy of some sensors, the weight of the virtual space data is appropriately increased to obtain the fused observation vector At the same time, monitor the interaction-related information between different parallel spaces, such as whether there is an intelligent driving induction space requesting the traffic data of this space.

[0157] Step (2) Evaluation phase: Calculate evaluation indicators, traffic congestion index Ascent, average travel time T t Variable length, MAE t The value may increase. t Exceeding the threshold value θ t , indicating that there is a large deviation between the virtual space simulation and the real space observation. Calculate the evaluation indicators of the interaction effect of different parallel spaces. If there is data exchange, calculate the accuracy of data exchange. If other spaces request the traffic data of this space, the accuracy and timeliness of the statistical data provided; the space conversion success rate is calculated based on whether there is a conversion request and whether the conversion is successful.

[0158] Step (3) Diagnosis phase: Trigger the diagnosis process and calculate the relative probability P of the virtual space model being inaccurate r The relative probability P of interference between (H1) and real space r (H2). If P r (H1) is large, check the virtual space state transfer function and the observation function Parameters, especially those related to traffic flow, vehicle speed, and queue length. At the same time, analyze whether the virtual space does not accurately consider factors such as vehicle interactions and driver behavior changes caused by congestion. If P r (H2) is relatively large. It is necessary to check whether there are interference factors such as traffic accidents and temporary road construction in the real space. This can be determined by analyzing the abnormal conditions of sensor data, the mutation points of traffic flow and vehicle speed, etc.

[0159] Step (4) Optimization phase: Optimize virtual space parameters based on the diagnosis results, such as adjusting the coefficients related to congestion in the traffic flow prediction model and adding parameters that reflect the impact of the length of the vehicle queue on the vehicle speed. At the same time, optimize the real space control strategy, such as adjusting the timing of traffic lights, increasing the green light duration in the congested direction, and setting a dynamic speed limit to relieve traffic pressure.

[0160] Step (5) Implementation phase: Apply the optimized control strategy to the real traffic system, such as adjusting the signal light control parameters and issuing new speed limit instructions. At the same time, update the virtual space state and continue the next round of data collection, evaluation, diagnosis and optimization.

[0161] Example 2: Bad weather space (such as )Implementation Example

[0162] Embodiment 2 of the present invention is a bad weather space (such as ), the specific implementation steps are as follows:

[0163] Step (1) Perception stage: In real space Ω R In the process, weather sensors detect severe weather information such as rain and snow, which affects traffic flow. Speed and traffic environment complexity Raindrop sensors (if available) can detect the size and density of raindrops to more accurately assess the impact of rainfall intensity on traffic. The camera may have reduced image clarity due to rain, fog, or snow, which may reduce its image recognition accuracy. This can be processed through image enhancement algorithms (but there may still be some errors). Geomagnetic sensors may increase traffic flow measurement errors due to factors such as water accumulation. The virtual space adjusts model parameters according to the type and intensity of severe weather, for example, considering the impact of rain on road friction to adjust the vehicle speed prediction model, and considering the impact of reduced visibility on traffic flow (visibility can be estimated through meteorological sensor data or camera image analysis). When fusing data, the weights are adjusted according to the reliability of different sensors in severe weather and the credibility of the virtual space in severe weather simulation. At the same time, monitor the interaction-related information between different parallel spaces, such as whether there is an intelligent driving guidance space requesting severe weather data for adjusting the guidance route.

[0164] Step (2) Evaluation phase: Calculate evaluation indicators. Severe weather causes significant changes in traffic congestion index, average travel time and other indicators. MAE t Fluctuations may increase. Based on these indicators, evaluate the simulation effect of virtual space on traffic conditions in bad weather. Calculate the evaluation indicators of the interaction effects of different parallel spaces. If there is data exchange, evaluate the degree of help of data exchange in traffic management in bad weather. For example, after the intelligent driving guidance space uses the bad weather space data to adjust the guidance route, it is statistically analyzed whether the road traffic efficiency is improved; the space conversion success rate is calculated based on whether the bad weather changes trigger the space conversion and whether the conversion is successful.

[0165] Step (3) Diagnosis stage: If MAE tIf the threshold is exceeded, the diagnosis is started. Analyze the model part related to bad weather in the virtual space, such as whether the influence function of weather factors on traffic parameters is accurate. Check the interference of bad weather on sensors in the real space, such as whether the rain sensor is working properly and whether the road humidity sensor data is accurate. At the same time, analyze whether there are special situations such as traffic accidents caused by bad weather and difficulty in vehicle passage caused by water accumulation on the road. If a sensor failure is found, record the fault information in time for subsequent maintenance or adjustment of the data fusion strategy. If there are traffic accidents or difficult road passage areas caused by bad weather, mark the relevant information and pass it to the control execution layer so that corresponding traffic control measures can be taken.

[0166] Step (4) Optimization phase: Optimize virtual space parameters, such as adjusting the weight of weather factors in the traffic flow prediction model, and improving the parameters related to road friction and visibility in the vehicle speed and energy consumption model. For the real space, adjust the control strategy, such as lowering the speed limit and extending the signal light cycle to adapt to traffic conditions in bad weather, and consider issuing special traffic guidance information, such as reminding drivers to pay attention to flooded sections and slippery sections. If there is interaction with the intelligent driving guidance space, further optimize the guidance strategy based on its feedback information, such as providing more accurate route planning to avoid flooded and congested sections.

[0167] Step (5) Implementation phase: Implement the optimized control strategy to ensure that the instructions are accurately transmitted to the traffic control equipment. Update the virtual space status, continuously monitor traffic conditions and weather changes, and prepare for the next round of closed-loop operations. During the implementation process, if the traffic control equipment responds abnormally, check the communication link and equipment status in time to ensure the effective execution of the control strategy. At the same time, feed back the updated virtual space status and real space observation data to other related parallel spaces (such as intelligent driving guidance space) so that they can make corresponding adjustments and optimizations.

[0168] Example 3: Traffic support space for major events (such as )Implementation Example

[0169] Embodiment 3 of the present invention is a major event traffic support space (such as ), the specific implementation steps are as follows:

[0170] Step (1) Perception stage: In the event preparation stage, data is shared with the event organizer to obtain relevant information about the event, such as the event type, scale, time, location, and expected number of participants. R Temporary sensors (such as vehicle counting sensors, pedestrian detection sensors, etc.) are added around the event site and related roads to strengthen the collection of data such as traffic flow, vehicle speed, and pedestrian flow. At the same time, weather information is received from meteorological sensors to cope with possible bad weather. In the process of data fusion, the real-time data of the real space and the simulated data of the virtual space are comprehensively considered to reasonably determine the fusion weight, and focus on the data accuracy of the activity-related areas. At the same time, the interaction needs with other spaces (such as intelligent driving guidance space, traffic congestion space, etc.) are monitored, such as providing activity traffic control information to the intelligent driving guidance space so that it can provide accurate guidance services to drivers; working with the traffic congestion space to prevent possible congestion in the roads around the activity in advance.

[0171] Step (2) Evaluation phase: Calculate evaluation indicators, including the traffic congestion index, average travel time, and parking resource utilization rate of roads around the event site, to evaluate the effectiveness of event traffic security measures. For example, calculate the utilization rate of parking lots around the event site to ensure that there are enough parking spaces to meet vehicle parking needs; monitor traffic congestion on roads during the event to avoid delays for event participants due to traffic congestion. At the same time, evaluate the impact of interaction with other parallel spaces on event traffic security, such as whether the guidance information provided by the intelligent driving guidance space effectively guides vehicles to the event site, and whether the congestion warning of the traffic congestion space is timely and accurate. Through these evaluation indicators, determine whether the event traffic management strategy needs to be adjusted.

[0172] Step (3) Diagnosis phase: If the evaluation indicators show that there are problems with traffic security, such as severe traffic congestion and parking difficulties, the diagnosis process is triggered. Analyze the accuracy of the virtual space activity traffic model to check whether the changes in the actual number of participants in the activity, the impact of temporary traffic control measures on site, and other factors are fully considered. At the same time, check the working status of temporary sensors in the real space to ensure that data collection is accurate. For example, if it is found that the utilization rate of a parking lot is far lower than expected, check whether the parking lot entrance sensor is working properly and whether the induction information accurately guides vehicles to the parking lot. In addition, analyze whether there is information stagnation or inadequate coordination in the interaction with other spaces, such as whether the induction route with the intelligent driving induction space is reasonably connected, and whether it is coordinated with the congestion relief strategy of the traffic congestion space.

[0173] Step (4) Optimization phase: Optimize virtual space parameters based on the diagnosis results, such as adjusting the traffic flow prediction parameters in the event traffic model, and optimizing the parking demand prediction model based on actual conditions. At the same time, optimize the real space control strategy, such as adjusting the signal light timing, setting up special lanes and priority rights of way for event-related vehicles, and reasonably guiding vehicle traffic and parking. For example, before the event starts, increase the green light duration of the main entrance road to the event venue and reduce the green light duration in other directions to ensure that event vehicles can enter the venue quickly; at the end of the event, dynamically adjust the signal light timing of the surrounding roads based on the departure of the parking lot vehicles to speed up the evacuation of vehicles. If there is interaction with other spaces, further optimize the traffic guarantee measures based on the interaction situation. For example, according to the feedback from the intelligent driving induction space, adjust the traffic control scope and induction information release strategy of the roads around the event site; coordinate with the traffic congestion space, share traffic diversion resources, and jointly deal with possible traffic congestion problems.

[0174] Step (5) Implementation phase: Implement the optimized traffic security strategy, ensure traffic order during the event through on-site command by traffic police, remote control of traffic control equipment (signal lights, variable speed limit signs, etc.), and collaboration with event organizers (such as adjusting event entry and exit times, etc.). At the same time, update the virtual space state, adjust the simulation results according to the actual traffic operation conditions and control strategies, continuously monitor the traffic conditions, and prepare for the next round of closed-loop operations. During the implementation process, pay close attention to changes in traffic conditions and handle emergencies in a timely manner, such as traffic accidents and vehicle failures, to ensure safe and smooth event traffic. If a traffic control device or temporary sensor fails, immediately activate the emergency plan, perform equipment maintenance or data backup and recovery, and ensure continuous collection of traffic data and effective execution of control strategies.

[0175] Example 4: Parallel Space Conversion Process Implementation Example (Based on Sunny Space To the rainy day space For example)

[0176] Embodiment 4 of the present invention is a sunny space To the rainy day space The conversion process (such as Figure 3 The specific implementation steps are as follows:

[0177] Step (1) Space conversion trigger: In real space Ω R In the process, the meteorological sensor detects the gradual increase in rainfall, and the raindrop sensor (if any) feedbacks the raindrop density and size information. At the same time, the camera captures the gradual appearance of water stains on the road surface, and image analysis shows that visibility is reduced. The speed sensor shows the speed of the vehicle. Begins to descend, traffic flow captured by geomagnetic sensors Data fluctuates and traffic environment becomes more complex Rising due to weather changes. In the virtual space In the simulation, the current data is based on the sunny model. As the real space data changes, the deviation from the real space observation value gradually increases. According to the conversion function Calculate the transition probability from sunny space to rainy space when At the same time, the spatial conversion request related information is recorded, including the time of triggering the conversion, the changes in key parameters (such as the amount of rainfall that triggers the conversion), etc., for subsequent analysis and optimization.

[0178] Step (2) Transformation execution: Adjustment of the perception layer: In real space, increase the acquisition frequency of meteorological sensors to obtain more accurate information such as rainfall amount and rainfall intensity. Adjust the image enhancement algorithm and parameters of the camera to improve the ability to identify vehicles and road conditions in rainy and foggy environments. For speed and traffic flow sensors, calibrate their possible errors under slippery road conditions. In virtual space Update the initial parameters of the virtual space. Adjust the vehicle speed simulation model and reset the friction coefficient related parameters according to the effect of rain on road friction. Modify the traffic flow model to consider the impact of reduced visibility on driver driving behavior and traffic flow, such as increasing the parameters affecting traffic flow due to factors such as increased driver reaction time and increased vehicle spacing. At the same time, adjust the traffic environment complexity model to include factors such as rainwater accumulation and road slipperiness. In addition, if there has been data exchange or collaboration with other spaces (such as intelligent driving induction space) before, promptly notify the relevant space of weather changes so that it can adjust its strategy.

[0179] Step (3) The data processing layer adjustment of the conversion execution is specifically implemented as follows:

[0180] Step (3.1) Data fusion: Redefine the fusion weight. Since the simulated data of weather-related factors in the virtual space become more important in rainy days, increase the weight of rainy-day related data in the virtual space (such as the simulated visibility impact on vehicle speed, the impact of slippery road surface on traffic flow). For example, for traffic flow data fusion, the new fusion formula contains the rainy-day space virtual space traffic flow data. Weight Increase and adjust real space traffic flow data The weight α Q,t .

[0181] Step (3.2) Evaluation index calculation: Add evaluation indicators for rainy day-specific factors. In addition to the original traffic congestion index, average travel time, and virtual-real space fitting degree indicators, new indicators such as the impact of road water depth on vehicle speed, wiper usage frequency (which can be estimated through vehicle-mounted sensor data, if there are relevant sensors) and the relationship between line of sight obstruction, etc., are added. These new indicators can more comprehensively evaluate rainy day traffic conditions and system performance. At the same time, adjust the weights of weather-related parameters in the calculation of evaluation indicators. For example, in the calculation of traffic congestion index, increase the weight of rainy weather on the complexity of the traffic environment.

[0182] Step (3.3) Diagnosis and optimization process: In the diagnosis phase, focus on problems caused by rain. Analyze the accuracy of rain-related parameters in the virtual space, such as whether the settings of the parameters affecting the braking distance of different types of vehicles in the speed model are reasonable. Check the working status of sensors in the real space on rainy days, such as whether the rain sensor accurately measures the rainfall and whether the road humidity sensor data is consistent with the actual water accumulation. During the optimization process, adjust the virtual space parameters for the diagnosed problems, such as further optimizing the parameters that consider road friction and water accumulation depth in the speed model, and adjusting the vehicle following model parameters in the traffic flow model due to reduced visibility. If there is a need for collaborative optimization with other spaces (such as traffic congestion space and intelligent driving induction space), adjust the strategy together according to the rainy day conditions, such as coordinating with the traffic congestion space to optimize the signal light timing to adapt to the changes in traffic flow on rainy days, and sharing rainy day road condition information with the intelligent driving induction space to optimize the induction route.

[0183] Step (4) Control execution layer adjustment of conversion execution: Adjust the control strategy according to the traffic characteristics of rainy days. Reduce the speed limit command. For example, on the highway, according to the rainfall and road water accumulation, appropriately reduce the speed limit value from 120km / h on sunny days to 80-100km / h. Adjust the timing of the signal light and increase the transition time of the green light duration to adapt to the situation that the driver's reaction speed is slowed down on rainy days. Publish more safety tips related to rainy days through the traffic guidance screen, such as "Please drive carefully in the flooded section ahead", "Pay attention to keep a safe distance to prevent braking in time", etc., and provide information on the surrounding sections prone to flooding and accident-prone sections. If the transition from sunny to rainy days occurs during the traffic guarantee period of major events, timely adjust the traffic control measures around the event site, such as adding anti-skid facilities, adjusting the layout of temporary parking lots to avoid water accumulation affecting vehicle parking, etc. For the intelligent driving guidance space, provide more detailed rainy road condition information to help drivers better plan routes.

[0184] Step (5) Model management layer adjustment for conversion execution: record from Sunny Space To the rainy day space The conversion information of the virtual and real space is collected, including the conversion time and the key parameter values ​​that trigger the conversion (such as rainfall, vehicle speed change, etc.). The virtual and real space model version is updated and the new model parameters are saved, such as the new parameters of the vehicle speed, traffic flow, and traffic environment complexity model of the rainy day space. According to the historical data of the rainy day space (if any) and the current operation status, the update strategy of the model parameters is further adjusted. For example, when the rainfall intensity changes beyond a certain threshold, the update frequency of the rainy day related parameters in the virtual space is accelerated to ensure the continuous accuracy and adaptability of the model in the rainy day space.

[0185] Step (6) Continuous monitoring and optimization (closed-loop operation in the new space): After the control execution layer implements the adjusted control strategy, the traffic conditions are continuously monitored. The real space continues to collect data through sensors and updates the real space observation vector And according to the real space observation equation, the real space traffic parameter vector is inferred The virtual space updates the virtual space state vector according to the new observation data and the updated model parameters and the observation vector According to the new real space observation vector and the virtual space observation vector Re-perform closed-loop operations such as data fusion, evaluation, diagnosis, and optimization. If new problems are found in the new round of closed-loop, such as a sudden increase in rainfall causing local waterlogging that seriously affects traffic, or deviations in the simulation of new traffic conditions in the virtual space, perform corresponding diagnosis and optimization operations again to continuously improve system performance and ensure the stable operation of the traffic system in rainy days.

[0186] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system, characterized in that: The method comprises the following steps: Step 1, divide multiple parallel traffic spaces and construct virtual and real space traffic parameter vectors; Step 2: construct an interactive model of multiple parallel traffic spaces; Step 3, modeling state equations and observation equations of multiple parallel virtual and real spaces; Step 4: Realize the iterative operation of the parallel traffic system through the closed-loop logic of "perception-assessment-diagnosis-optimization-implementation".

2. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 1, characterized in that: In step 1, multiple parallel traffic spaces are divided, including: Based on traffic flow, speed, vehicle density, weather conditions, road type and functional requirements, the traffic system is divided into a physical real space Ω R and multiple parallel digital virtual spaces n is the number of virtual spaces; The traffic system division includes dividing into free space according to the degree of traffic congestion Lightly congested space Severely congested space The traffic system division also includes clear sky space according to weather conditions Rainy Day Space Snow Space The traffic system division also includes intelligent driving guidance function space according to functional requirements Functional space for transportation support of major events 3. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 1, characterized in that: In step 1, the virtual-real space traffic parameter vector is constructed, which specifically includes: Real space traffic parameter construction: real space Ω R The traffic parameter vector at time t is in is the real space Ω R The traffic flow at time t is is the real space Ω R The vehicle speed at time t, is the real space Ω R The complexity of the traffic environment at time t; Construction of virtual space traffic parameters: virtual space The traffic parameter vector at time t is in Virtual Space The traffic flow at time t is Virtual Space The vehicle speed at time t, Virtual Space The complexity of the traffic environment at time t.

4. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 1, characterized in that: Step 2 describes constructing an interactive model of multiple parallel traffic spaces, specifically including: Definition from A i Virtual Space To A j Virtual Space The conversion function Among them, ω1, ω2, ω3 and ω4 are weight coefficients used to adjust the importance of each parameter in the conversion decision. is the real space Ω r The traffic parameter vector at time T, It is virtual space The traffic parameter vector at time T, is the control quantity of real space at time T; It is virtual space The simulated control quantity is generated according to the virtual space model and control strategy; The interaction methods between different parallel spaces include: (1) Data exchange: Data is exchanged between different parallel traffic spaces based on demand; (2) Spatial transformation: When the transformation function satisfies , spatial transformation is performed, and the transformation process involves the adjustment of the perception layer, data processing layer, control execution layer and model management layer; where θ ij The conversion threshold is pre-set according to different space conversion scenarios and traffic management requirements; (3) Collaborative iteration: Multiple parallel traffic spaces work together to perform iterative optimization.

5. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 1, characterized in that: The state equations and observation equations for modeling multiple parallel virtual and real spaces described in step 3 specifically include: Step 3-1, establish the state equation of the virtual space at time t: In the formula, is the virtual space at time t state vector, is the virtual space at time t Simulation credibility index, used to measure the reliability of virtual space simulation results; is the state transfer function based on the preset model at time t, Virtual Space Current Status and observational data Map to the state of the next moment; is the process noise at time t; Step 3-2, establish the observation equation of virtual space at time t: In the formula, is the observation vector, They correspond to the observed values ​​of traffic flow, vehicle speed, and traffic environment complexity respectively; is the observation function, which is used to transform the virtual space state vector Mapping to observation space; is the observation noise; Step 3-3, observation equation of real space at time t: In the formula, is the real space observation vector, They correspond to the observed values ​​of traffic flow, vehicle speed, and traffic environment complexity, respectively. is the real space observation function; is the real-space observation noise.

6. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 1, characterized in that: Step 4 implements the iterative operation of the parallel traffic system through the closed-loop logic of "perception-assessment-diagnosis-optimization-implementation", which specifically includes: Step 4-1, perception phase: collect and fuse real space and virtual space data to provide a data basis for subsequent operations, while monitoring information related to interactions between different parallel spaces; Step 4-2, evaluation phase: Calculate multiple evaluation indicators and conduct spatial evaluation to determine the virtual space simulation effect, and evaluate the impact of different parallel space interactions on the transportation system; Step 4-3, diagnosis phase: trigger diagnosis based on specific thresholds to analyze the source of virtual space or real space problems; Step 4-4, optimization phase: optimize virtual space parameters and real space control strategy according to the diagnosis results; Step 4-5, implementation phase: Apply optimization strategies, update the real and virtual space states and observation vectors, prepare for the next round of closed loop, and ensure the effective execution of interactions between different parallel spaces.

7. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 6, characterized in that: Step 4-1 specifically includes: Step 4-1-1, data collection: In real space Ω R In the process, traffic parameter data is collected through various sensors to obtain the real space observation vector In virtual space In the process, simulated traffic parameter data is generated according to the preset model and algorithm, and combined with some real space observation data to obtain the virtual space state vector and the observation vector Step 4-1-2, data fusion: fuse the traffic parameter data.

8. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 6, characterized in that: Step 4-2 specifically includes: Step 4-2-1, calculation of evaluation indicators, wherein the evaluation indicators include traffic congestion index, average travel time, virtual-real space fitting index, accuracy of data exchange between different parallel spaces, and space conversion success rate; Step 4-2-2, space evaluation based on evaluation indicators: According to the evaluation indicators calculated in step 4-2-1, compare the differences in evaluation indicators between different virtual spaces and real spaces, and evaluate the simulation effect of virtual space on real space through these differences.

9. The method for realizing a virtual-reality interactive and closed-loop driven parallel traffic system according to claim 6, characterized in that: Step 4-3 specifically includes: Step 4-3-1, diagnostic trigger mechanism setting, when MAE t >θ t When the diagnostic process is triggered, MAE t represents the fitting degree index of virtual and real space at time t, θ t It is a threshold that is dynamically adjusted according to different traffic scenarios and system performance requirements; Step 4-3-2, Problem source analysis: Calculate the relative probability P of the virtual space model being inaccurate r The relative probability P of interference between (H1) and real space r (H2) = 1-P r (H1), according to P r (H1) and P r (H2) can be used to initially determine the source of the problem. If P r (H1) is large, it indicates that the virtual space model may be inaccurate; if P r If (H2) is large, it indicates that there may be interference in the real space; Step 4-3-3, further analysis and adjustment: If the virtual space model is inaccurate, diagnose the structure of the virtual space model to determine whether the current structure of the virtual space model is reasonable. If not, make adaptive adjustments.

10. The method for realizing a virtual-real interaction and closed-loop driven parallel traffic system according to claim 6, characterized in that: Step 4-4 specifically includes: Step 4-4-1, virtual space parameter optimization: Use gradient-based optimization algorithm to optimize virtual space parameters To minimize the virtual-real space fitting degree index MAE t is the objective function; Step 4-4-2, real space control strategy optimization: adjust the real space control strategy according to the virtual space parameter optimization results Steps 4-5 specifically include: Step 4-5-1, control strategy implementation: Apply the optimized real space control strategy Applied to real traffic system; Step 4-5-2, system status update: real space status update, according to the actual operation of the real space, re-collect data through the sensor, and update the real space observation vector at the next moment And according to the real space observation equation, the real space traffic parameter vector of the next moment is inferred Step 4-5-3, preparation for a new round of closed-loop operation: the new real-space observation vector obtained in step 4-5-2 and the virtual space observation vector Re-execute the closed-loop operation from step 4-1 to step 4-4 to enable the system to continuously adapt to the dynamic changes in the traffic environment and continuously optimize system performance.

Images