A method for traffic emergence identification and new feature discovery in IVCPS

Abstract

This invention belongs to the field of intelligent connected vehicles and vehicle-road-cloud integration technology, and discloses a method for traffic emergence recognition and new function discovery for IVCPS. This invention collects multi-dimensional traffic states, constructs univariate and joint probability densities, and achieves early and accurate recognition of traffic emergence based on distribution distance; then, it constructs a novelty index through mutual information to quantify the novelty of system functions; finally, it uses novelty as a reward-guided parameter to automatically discover new functions. This invention can effectively capture early emergence, quantify new functions, and form an integrated closed loop of recognition and discovery, improving the autonomous cognition and proactive optimization capabilities of IVCPS, and is applicable to intelligent road networks, vehicle-road cooperation, and proactive traffic management.

Description

Technical Field

[0001] This invention belongs to the field of intelligent connected vehicles and vehicle-road-cloud integration technology, specifically involving a method for traffic emergence recognition and new function discovery for IVCPS. Background Technology

[0002] In urban transportation systems where Vehicle-Road-Cloud Integrated Systems (IVCPS) are gradually being implemented, traffic operations are jointly formed by the interaction of multiple stakeholders—people, vehicles, roads, edges, and the cloud—and are characterized by strong nonlinearity, strong coupling, and cross-scale propagation. Traffic "emergence" typically manifests as the rapid evolution of traffic conditions from localized micro-level changes to meso- / macro-level structural changes (queue wave diffusion, sudden capacity drop, sudden risk increase, abrupt changes in collaborative benefits, etc.) under the influence of disturbances (accidents, sudden demand surges, signal mismatches, platoon disturbance propagation, etc.) or changes in system conditions (connected vehicle penetration rate, collaborative strategies, control parameters, communication and computing power conditions, etc.). Simultaneously, with the enhancement of vehicle-road collaborative capabilities, the system may also generate "new functions" under specific conditions, manifested as the triggering of new collaborative mechanisms or control laws, leading to a leapfrog improvement in traffic efficiency, stability, or safety.

[0003] Currently, the main methods for traffic emergence recognition and new feature discovery are as follows:

[0004] (1) Emergent / anomaly identification methods based on single-index thresholds and rule bases. These methods typically use indicators such as speed, flow, occupancy, queue length, and travel time, combined with fixed thresholds or empirical rules, to identify congestion, events, or anomalies. Their advantages are simple implementation, low computational overhead, and ease of engineering deployment. Their disadvantages are: they mainly rely on the mean or simple statistics, making it difficult to identify early distribution pattern changes and variable coupling structure reorganization; they are also sensitive to changes in road segments, time periods, and traffic demand, making them prone to false alarms and false negatives; and they are also difficult to provide effective clues for subsequent new function discovery.

[0005] (2) Emergent identification methods based on mechanistic models and state estimation. These methods rely on queuing models, cellular transmission models, macroscopic basic graphs, or state-space models to identify anomalies and mutations through state estimation, residual analysis, or mode switching. Their advantage lies in their interpretability and ease of integration with traffic control models. Their disadvantages include: the models typically rely on strong assumptions and parameter calibration, making them prone to bias under conditions of mixed traffic, vehicle-road cooperation, and multi-source heterogeneous data; limited adaptability to unknown emergent forms and complex interactions; and high model maintenance costs during long-term operation.

[0006] (3) Emergent identification methods based on data-driven / machine learning: These methods identify traffic mutations by using anomaly detection models, prediction models, or deep learning models, utilizing prediction errors, reconstruction errors, or anomaly scores. Their advantages lie in their ability to fit complex nonlinearities and to integrate multi-source data to improve identification accuracy. Their disadvantages include: most methods focus on error or scoring criteria, lacking a structured characterization of the emergent nature, making it difficult to explain how distributions evolve and how variable relationships are reorganized. They also face problems such as concept drift, model degradation, and continuous retraining, and are difficult to integrate with the discovery of emerging functions.

[0007] (4) Existing methods for discovering new features mainly rely on expert experience in parameter tuning, simulation comparison experiments, or performance optimization using methods such as reinforcement learning, evolutionary algorithms, and Bayesian optimization. The advantage is that it can automatically obtain usable strategies to a certain extent. The disadvantages are that: the search range of experience and enumeration methods is limited and the cost is high; pure performance-oriented optimization is prone to focus on known high-performance areas and lacks an explicit measure of "functional novelty". Therefore, even if a better strategy is obtained, it is difficult to determine whether it corresponds to a truly reproducible and interpretable new feature. Summary of the Invention

[0008] In view of this, the purpose of this invention is to provide a traffic emergence identification and new function discovery method for IVCPS, which aims to solve the technical problems of insufficient early sensitivity, inadequate characterization of coupling structure, poor interpretability, lack of novelty quantification in new function discovery, pursuit of optimal performance only, and inability to automatically discover reproducible new mechanisms, so as to achieve accurate early identification of traffic emergence and efficient automatic discovery of collaborative new functions.

[0009] To achieve the above objectives, the present invention adopts the following technical solution:

[0010] A traffic emergence identification and new feature discovery method for IVCPS includes two parts: a traffic emergence identification measurement method and a new feature discovery method.

[0011] The traffic emergence identification and measurement method uses traffic density ρ, vehicle speed v, and vehicle acceleration a as the core variables of traffic state. By constructing a joint probability density function of single variables and three variables, it defines an emergence intensity index to achieve early identification, discrimination, and localization of traffic emergence.

[0012] The newly discovered function discovery method defines the statistical dependency change between control / configuration parameters, emergence features, and performance output as the emergence index. This index is embedded as a reward in the search strategy to automatically discover high-value vehicle-road cooperative emerging function schemes within the control / configuration parameter space.

[0013] Furthermore, the traffic emergence identification and measurement method includes the following sub-steps:

[0014] S1.1 Variable Definition and Sample Construction:

[0015] The urban road network is divided into several road segments / areas, denoted as... Set a sliding time window The adjacent window step size is ; Index for sliding time windows;

[0016] In road sections / areas window The internal sampling yielded n sample triples:

[0017]

[0018] in, Indicates traffic density. , For a moment Section The number of vehicles inside, For road section Length; Indicates traffic flow speed. For a moment Section The average speed of all vehicles in the area; The vehicle acceleration is represented by a velocity difference calculation; sliding window. The sample set within is denoted as :

[0019]

[0020] in, Represents a sliding window The total number of samples collected internally;

[0021] S1.2 Construction of the joint probability density equation: Univariate probability density functions for ρ, v, and a are constructed using kernel density estimation, and then... Establish the joint probability density function of the three variables;

[0022] S1.3 Emergent Intensity Definition: Select a reference window based on normal operating conditions or historical stable periods. Using the corresponding density function as a reference, we define the univariate emergence intensities of ρ, v, and a, and the joint structure emergence intensities of ρ-va, respectively, and then obtain the comprehensive emergence index through weighted summation. ;

[0023] S1.4 Emergence Judgment Rules and Positioning Methods: The judgment threshold of the comprehensive emergence index is obtained based on the statistical data of the baseline period. ,when And when M consecutive windows are satisfied, the road segment / region is determined. In the window Traffic surges occur within the system; simultaneously, the contributions of variables in univariate structures ρ, v, and a, as well as in the ρ-va joint structure, are calculated. , , , Based on the contribution ratio analysis, the reasons for the emergence of the stage are as follows: When the proportion increases significantly, it is determined that the emergence is caused by the reorganization of the ρ-va coupling structure.

[0024] Furthermore, in step S1.2, the formulas for constructing the univariate probability density functions of ρ, v, and a are as follows:

[0025]

[0026]

[0027]

[0028] in, Represents road segment / area s, window Internal traffic density The probability density function, where r is the value of the density; Represents road segment / area s, window The probability density function of the internal traffic flow speed v, where u is the speed value; Represents road segment / area s, window Internal vehicle acceleration The probability density function, The value is taken for acceleration; They are respectively v Bandwidth for kernel density estimation; For kernel functions;

[0029] The formula for constructing the joint probability density function of the three variables is:

[0030]

[0031] in, Represents road segment / area s, window Inside ,v, Joint probability density function of three-dimensional variables; H is the bandwidth matrix.

[0032] Furthermore, step S1.3 includes the following sub-steps:

[0033] I. Define the univariate emergence intensity of ρ, v, and a;

[0034]

[0035]

[0036]

[0037] in, Represents road segment / area s, window Internal traffic density univariate emergence intensity; Represents road segment / area s, window Univariate emergence intensity of internal traffic flow velocity v; Represents road segment / area s, window Internal vehicle acceleration univariate emergence intensity; , , ρ, v, a are the similarity between the probability densities of the current window and the baseline window, respectively.

[0038] II. Define the emergent intensity of the ρ-va joint structure;

[0039]

[0040] in, Represents road segment / area s, window Emergent intensity of the inner ρ-va joint structure; Reference window for road segment / area s The ρ-va joint probability density function; For the current window of road segment / area s The ρ-va joint probability density function;

[0041] III. Calculate the comprehensive emergence index ;

[0042]

[0043] in, Let be the weighting coefficients for each emergence intensity term, satisfying the normalization constraint: .

[0044] Furthermore, in step S1.4, the formulas for calculating the variable contributions of univariate ρ, v, a and the ρ-va joint structure are as follows:

[0045] ,

[0046] , .

[0047] Furthermore, the method for discovering new functionalities includes the following sub-steps:

[0048] S2.1 Construction of the representation object and data set for new functions;

[0049] Define control / configuration parameter vectors The control / configuration parameters include one or more of the following: signal timing, speed guidance, formation threshold, and cloud-edge offloading threshold; for each group Run simulations or online tests to obtain performance indicators. and emergent eigenvectors To form a sample set ;

[0050] S2.2 Probability Distribution Construction: Let Constructing through kernel density estimation and joint probability density function and get synchronously marginal distribution and marginal distribution ;

[0051]

[0052] In the formula, Represents the sample set The total number of samples; Represents the parameter space kernel function; Represents the kernel function of the combined feature space; , These are the parameter vector and combined feature vector corresponding to the i-th sample group, respectively;

[0053] S2.3 Newborn Quantification Definition: Using Mutual Information To characterize the strength of the dependency between control / configuration parameters and emergent features and performance output, the normal baseline mutual information estimated from historical stable data or the initial policy set is set as... Then the degree of newness is defined as ;

[0054] S2.4 Search and discovery strategy based on newbornness reward: Normalized newbornness is embedded as a reward item into the node evaluation function of the search tree. Through the iterative process of node expansion, indicator extraction, newbornness calculation, and node score selection, control / configuration parameter schemes that combine high performance and high newbornness are discovered.

[0055] Furthermore, in step 2.4, the comprehensive scoring function for search tree node j is defined as:

[0056]

[0057] in, Average performance return for nodes; This represents the number of visits to the parent node. For nodes Number of visits; To explore coefficients; For nodes The corresponding normalized newborn degree; This is the weighting factor for newbornness.

[0058] Beneficial effects:

[0059] 1. This invention focuses on the probability density and joint coupling structure evolution of traffic state variables for emergence identification. Compared with traditional single-index thresholds, mechanism models and conventional data-driven methods, it can capture emergence precursors earlier, accurately identify traffic emergence caused by multi-variable coupling and recombination, effectively reduce false alarm rate and false negative rate, and improve identification sensitivity and reliability.

[0060] 2. This invention can simultaneously output the univariate emergence intensity, the joint structural emergence intensity, and the variable contribution, clearly distinguishing between ordinary fluctuations and structural emergence, realizing the quantifiable discrimination, localizable analysis, and interpretable characterization of emergence, and providing a clear basis for traffic control.

[0061] 3. This invention does not rely on strong mechanistic assumptions or manual experience rules, making it more adaptable to mixed traffic, vehicle-road cooperation, and multi-source heterogeneous data scenarios. It also has low model maintenance costs and can operate stably in IVCPS systems for a long time.

[0062] 4. This invention transforms emerging features into novelty rewards, enabling explicit quantification of functional novelty. It solves the problem that traditional methods only pursue optimal performance and cannot identify truly novel functions, and can automatically discover reproducible and interpretable traffic cooperative novel functions.

[0063] 5. This invention adopts a novelty-guided intelligent search strategy, which balances performance optimization and novelty exploration, significantly reduces the scope of invalid searches, lowers simulation and testing costs, and improves the efficiency and success rate of discovering new features.

[0064] 6. This invention forms a complete closed loop of emergence recognition, new function discovery, and control parameter output, directly outputting deployable signal timing, speed guidance, formation threshold, and other control parameters, adapting to the IVCPS "human-vehicle-road-edge-cloud" collaborative management and control requirements, and significantly improving system traffic efficiency, stability, and security.

[0065] Other advantages, objectives, and features of the invention will be set forth in part in the description which follows, and in part will be apparent to those skilled in the art from the following examination, or may be learned from practice of the invention. The objectives and other advantages of the invention can be realized and obtained through the following description. Attached Figure Description

[0066] Figure 1 This is a technical roadmap for a traffic emergence recognition and new function discovery method for IVCPS according to the present invention. Detailed Implementation

[0067] To make the technical solutions, advantages, and objectives of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the described embodiments of the present invention without creative effort are within the protection scope of this application.

[0068] like Figure 1 As shown, this invention provides a traffic emergence recognition and new feature discovery method for IVCPS, which includes two parts: a traffic emergence recognition measurement method and a new feature discovery method.

[0069] The traffic emergence identification and measurement method uses traffic density ρ, vehicle speed v, and vehicle acceleration a as the core variables of traffic state. By constructing a joint probability density function of univariate and trivariate variables, it defines an emergence intensity index to achieve early identification, discrimination, and localization of traffic emergence.

[0070] The emerging function discovery method defines the statistical dependency change between control / configuration parameters and emergent features and performance output as the emergence index. This index is embedded as a reward in the search strategy to automatically discover high-value vehicle-road cooperative emerging function schemes within the control / configuration parameter space.

[0071] 1. Traffic Emergence Recognition Measurement Method

[0072] 1.1 Variable Definition and Sample Construction

[0073] The urban road network is divided into several road segments / areas, denoted as... Set a sliding time window The adjacent window step size is ; Index for the sliding time window.

[0074] In road sections / areas window The internal sampling yielded n sample triples:

[0075]

[0076] Traffic density:

[0077]

[0078] Traffic speed:

[0079]

[0080] Vehicle acceleration (which can be obtained from velocity difference):

[0081]

[0082] in, Indicates traffic density; For road section Length; Indicates traffic flow speed. For a moment Section The average speed of all vehicles in the area; This represents the vehicle's acceleration, calculated using velocity difference. For a moment Located on the road section A collection of vehicles; sliding window The sample set within is denoted as :

[0083]

[0084] in, Represents a sliding window The total number of samples collected internally.

[0085] 1.2 Construction of Joint Probability Density Equation

[0086] (1) Constructing a univariate probability density equation using kernel density estimation

[0087] Construct density, velocity, and acceleration in the window respectively. The probability density function within:

[0088]

[0089]

[0090]

[0091] Among them, among them, Represents road segment / area s, window Internal traffic density The probability density function, where r is the value of the density; Represents road segment / area s, window The probability density function of the internal traffic flow speed v, where u is the speed value; Represents road segment / area s, window Internal vehicle acceleration The probability density function, The value is taken for acceleration; They are respectively v Bandwidth for kernel density estimation; For kernel functions;

[0092] (2) Construction of the joint probability density equation of the three variables

[0093] make Establish the joint probability density:

[0094]

[0095] in, Represents road segment / area s, window Inside ,v, The joint probability density function of the three-dimensional variables; H is the bandwidth matrix (a diagonal matrix can be used to reduce computational complexity).

[0096] 1.3 Definition of Emergence Intensity

[0097] Select the reference window Using the density function corresponding to (normal operating conditions or historical stable periods) as a reference, the emergence intensity is defined as "the difference between the current window distribution and the benchmark distribution".

[0098] (1) Univariate emergence intensity

[0099]

[0100]

[0101]

[0102] in, Represents road segment / area s, window Internal traffic density univariate emergence intensity; Represents road segment / area s, window Univariate emergence intensity of internal traffic flow velocity v; Represents road segment / area s, window Internal vehicle acceleration univariate emergence intensity; , , ρ, v, and a represent the similarity between the probability densities of the current window and the baseline window, respectively.

[0103] (2) Emergent intensity of joint structure

[0104]

[0105] in, Represents road segment / area s, window Emergent intensity of the inner ρ-va joint structure; Reference window for road segment / area s The ρ-va joint probability density function; For the current window of road segment / area s The ρ-va joint probability density function.

[0106] (3) Comprehensive Emergence Indicators

[0107]

[0108] in .

[0109] 1.4 Emergent discrimination rules and localization methods

[0110] (1) Threshold discrimination

[0111] The threshold is obtained based on statistics from the baseline period:

[0112]

[0113] When satisfied If the conditions are met consecutively for M windows, then the road segment / area is determined. In the window Emergence occurs.

[0114] (2) Variable contribution

[0115]

[0116]

[0117] when When the proportion increases significantly, it indicates that the emergence is mainly caused by the reorganization of the ρ-va coupling structure, rather than ordinary fluctuations of a single variable.

[0118] 2. Methods for Discovering New Functions

[0119] 2.1 Representation Objects and Data Sets for New Functionality

[0120] Define control / configuration parameter vectors (which can correspond to signal timing, speed guidance, formation thresholds, cloud-edge offloading thresholds, etc.):

[0121]

[0122] For each group Run simulations or online tests to obtain performance indicators:

[0123]

[0124] (For example: average delay, throughput, queue length, conflict risk indicators, etc.)

[0125] Simultaneously extract emergent feature vectors:

[0126]

[0127] Forming a sample set:

[0128]

[0129] 2.2 Construction of Probability Distribution

[0130] make Constructing through kernel density estimation and joint probability density function and get synchronously marginal distribution and marginal distribution ;

[0131]

[0132] In the formula, Represents the sample set The total number of samples; Represents the parameter space kernel function; Represents the kernel function of the combined feature space; , These are the parameter vector and combined feature vector corresponding to the i-th sample group, respectively.

[0133] 2.3 Quantitative Definition of Novelty (Functional Novelty)

[0134] Mutual information is defined to characterize the strength of the dependency between control parameters, emergent features, and performance output:

[0135]

[0136] Let the normal baseline mutual information be (Estimated from historical stable data or the initial policy set), then the degree of novicehood is defined as:

[0137]

[0138] when When the value is large, it indicates that changes in certain parameters have triggered a new "emergent structure-performance" correlation mechanism, meaning that there are new potential functional candidates.

[0139] 2.4 Search and Discovery Strategy Based on Novelty Reward

[0140] This invention employs a node selection strategy with a new node reward for searching.

[0141] (1) Node evaluation function (improved selection strategy)

[0142] For search tree nodes Define the overall score:

[0143]

[0144] in, For the average performance return of a node (e.g.) or ); This represents the number of visits to the parent node. For nodes Number of visits; To explore coefficients; For nodes The corresponding normalized newborn degree; This is the weighting factor for newbornness.

[0145] (2) Newborn Function Search Steps

[0146] Step 1: Initialization, setting control / configuration parameters: search space boundaries, initial node for tree search, performance metric calculation method, and sliding window length. With step size ;

[0147] Step 2: Generate candidate parameter groups, and generate candidate control / configuration parameters according to the node expansion strategy. And perform simulations or online experiments on the candidate parameter groups;

[0148] Step 3: Extract emergent features and performance metrics, and calculate the emergent feature vectors corresponding to the candidate parameter groups based on the traffic emergence recognition metric method. With performance indicators ;

[0149] Step 4: Calculate the newborn reward and update the joint probability density function based on the sample set. And calculate mutual information The newbornness corresponding to the candidate parameter set is obtained. ;

[0150] Step 5: Node scoring and selection. Calculate the comprehensive score for each node based on the comprehensive scoring function. Select the node with the highest score to continue expanding, and at the same time send the performance report back to update the relevant parameters of the node;

[0151] Step 6: Termination and Output. When the preset number of iterations or convergence condition is reached, output a high-performance and highly novel candidate control / configuration parameter scheme. The algorithm outputs the corresponding emergent feature explanation, which includes the contribution of each variable and the change of the ρ-va joint term.

[0152] It is hereby declared that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for traffic emergence identification and new feature discovery in IVCPS, characterized in that: It includes two parts: a traffic emergence identification and measurement method and a new function discovery method; The traffic emergence identification and measurement method uses traffic density ρ, vehicle speed v, and vehicle acceleration a as the core variables of traffic state. By constructing a joint probability density function of single variables and three variables, it defines an emergence intensity index to achieve early identification, discrimination, and localization of traffic emergence. The newly discovered function discovery method defines the statistical dependency change between control / configuration parameters, emergence features, and performance output as the emergence index. This index is embedded as a reward in the search strategy to automatically discover high-value vehicle-road cooperative emerging function schemes within the control / configuration parameter space.

2. The method for traffic emergence recognition and new function discovery for IVCPS according to claim 1, characterized in that, The traffic emergence identification and measurement method includes the following sub-steps: S1.1 Variable Definition and Sample Construction: The urban road network is divided into several road segments / areas, denoted as... Set a sliding time window The adjacent window step size is ; Index for sliding time windows; In road sections / areas window The internal sampling yielded n sample triples: in, Indicates traffic density. , For a moment Section The number of vehicles inside, For road section Length; Indicates traffic flow speed. For a moment Section The average speed of all vehicles in the area; The vehicle acceleration is represented by a velocity difference calculation; sliding window. The sample set within is denoted as : in, Represents a sliding window The total number of samples collected internally; S1.2 Construction of the Joint Probability Density Equation: Univariate probability density functions for ρ, v, and a are constructed using kernel density estimation, and then... Establish the joint probability density function of the three variables; S1.3 Emergent Intensity Definition: Select a reference window based on normal operating conditions or historical stable periods. Using the corresponding density function as a reference, we define the univariate emergence intensities of ρ, v, and a, and the joint structure emergence intensities of ρ-va, respectively, and then obtain the comprehensive emergence index through weighted summation. ; S1.4 Emergence Judgment Rules and Positioning Methods: The judgment threshold of the comprehensive emergence index is obtained based on the statistical data of the baseline period. ,when And when M consecutive windows are satisfied, the road segment / region is determined. In the window Traffic surges occur within the system; simultaneously, the contributions of variables in univariate structures ρ, v, and a, as well as in the ρ-va joint structure, are calculated. , , , The reasons for the emergence of the emerging stage are analyzed based on the contribution ratio.

3. The method for constructing complex environment maps based on end-to-end residual triggering according to claim 2, characterized in that, In step S1.2, the formulas for constructing the univariate probability density functions of ρ, v, and a are as follows: in, Represents road segment / area s, window Internal traffic density The probability density function, where r is the value of the density; Represents road segment / area s, window The probability density function of the internal traffic flow speed v, where u is the speed value; Represents road segment / area s, window Internal vehicle acceleration The probability density function, The value is taken for acceleration; They are respectively v Bandwidth for kernel density estimation; For kernel functions; The formula for constructing the joint probability density function of the three variables is: in, Represents road segment / area s, window Inside ,v, Joint probability density function of three-dimensional variables; H is the bandwidth matrix.

4. The method for traffic emergence identification and new function discovery for IVCPS according to claim 3, characterized in that, Step S1.3 includes the following sub-steps: I. Define the univariate emergence intensity of ρ, v, and a; in, Represents road segment / area s, window Internal traffic density univariate emergence intensity; Represents road segment / area s, window Univariate emergence intensity of internal traffic flow velocity v; Represents road segment / area s, window Internal vehicle acceleration univariate emergence intensity; , , ρ, v, a are the similarity between the probability densities of the current window and the baseline window, respectively. II. Define the emergent intensity of the ρ-va joint structure; in, Represents road segment / area s, window Emergent intensity of the inner ρ-va joint structure; Reference window for road segment / area s The ρ-va joint probability density function; For the current window of road segment / area s The ρ-va joint probability density function; III. Calculate the comprehensive emergence index ; in, Let be the weighting coefficients for each emergence intensity term, satisfying the normalization constraint: .

5. A method for traffic emergence identification and new function discovery for IVCPS according to claim 4, characterized in that, In step S1.4, the formulas for calculating the variable contributions of univariate ρ, v, a and ρ-va joint structures are as follows: ， ， 。 6. The method for traffic emergence recognition and new function discovery for IVCPS according to claim 5, characterized in that, The method for discovering new functionalities includes the following sub-steps: S2.1 Construction of the representation object and data set for new functions; Define control / configuration parameter vectors The control / configuration parameters include one or more of the following: signal timing, speed guidance, formation threshold, and cloud-edge offloading threshold; for each group Run simulations or online tests to obtain performance indicators. and emergent eigenvectors To form a sample set ; S2.2 Probability Distribution Construction: Let Constructing through kernel density estimation and joint probability density function and get simultaneously marginal distribution and marginal distribution ; In the formula, Represents the sample set The total number of samples; Represents the parameter space kernel function; Represents the kernel function of the combined feature space; , These are the parameter vector and combined feature vector corresponding to the i-th sample group, respectively; S2.3 Newborn Quantification Definition: Using Mutual Information To characterize the strength of the dependency between control / configuration parameters and emergent features and performance output, the normal baseline mutual information estimated from historical stable data or the initial policy set is set as... Then the degree of newness is defined as ; S2.4 Search and discovery strategy based on newbornness reward: Normalized newbornness is embedded as a reward item into the node evaluation function of the search tree. Through the iterative process of node expansion, indicator extraction, newbornness calculation, and node score selection, control / configuration parameter schemes that combine high performance and high newbornness are discovered.

7. A method for traffic emergence identification and new function discovery for IVCPS according to claim 6, characterized in that, In step 2.4, the comprehensive scoring function for search tree node j is defined as: in, Average performance return for nodes; This represents the number of times the parent node has been visited. For nodes Number of visits; For the exploration coefficient; For nodes The corresponding normalized newborn degree; This is the weighting factor for newbornness.

Images