CN-117351715-B - Vehicle path forecasting method based on generalized likelihood Kripke model

Abstract

A vehicle path forecasting method based on a generalized likelihood Kripke structural model comprises the steps of constructing an actual road condition model, constructing a path transfer distribution matrix, quantifying initial traffic conditions, constructing a vehicle path forecasting atomic proposition, quantifying a label value, carrying out model detection, evaluating the quality of a road intersection and forecasting an optimal path. Compared with a potential Kripke structure, the method for forecasting the optimal path by using the Kripke model has the advantages that a generalized potential Kripke model is adopted to model the road traffic condition, a road intersection is used as a state space, the smoothness degree of a road section is used as a path transfer distribution matrix, the average passing speed of a vehicle, the average waiting time of a red light and the traffic violation index are used as tag values, the optimal path forecasting method based on the modeling of the Kripke structure, the technical problem of difficult calculation is solved, the method has the advantages of easiness in data acquisition, simplicity and convenience in executing process and the like, and the method has a certain value for implementation measures of traffic management and forecasting.

Inventors

• LI YONGMING
• SU YIFAN
• HE QING
• LIU WUNIU

Assignees

• 陕西师范大学

Dates

Publication Date

20260512

Application Date

20231011

Claims (2)

1. 1. A vehicle path forecasting method based on a generalized likelihood Kripke model is characterized by comprising the following steps: (1) Constructing an actual road condition model Collecting road condition data between a starting point and a destination, and collecting all passing road intersections S between the starting point and the destination: S= Wherein, the Any element in the set S is represented, and n represents the number of road intersections; (2) Constructing a path transfer distribution matrix The path transfer distribution matrix P is constructed as follows: Wherein s m and s n represent road intersections, Representing the path-clearance index between road intersections s m and s n , ; (3) Quantifying initial traffic conditions The initial traffic condition at road intersection s n is quantized to column vector I n as follows: , Wherein i n is 1, and the rest elements are 0, which means that the first road intersection through which the vehicle passes is s n ; (4) Construction of atomic propositions for vehicle path forecast The vehicle path planning atomic proposition comprises a vehicle average speed v, a traffic light average waiting time t and a traffic violation index p; (5) Quantizing the tag value Quantifying vehicle average passing speed label value as follows : Wherein, the Representing the average passing speed of the vehicle, v max representing the maximum speed of all vehicles passing through the road intersection s n in the last half year, v min representing the minimum speed of all vehicles passing through the road intersection s n in the last half year, and the average waiting time tag value of the red light is quantized as follows : Wherein t represents the average waiting time of the red light, t max represents the longest waiting time of all vehicles passing through the road intersection s n in the last half year, t min represents the shortest waiting time of all vehicles passing through the road intersection s n in the last half year, and the traffic violation index tag value is quantified according to the following formula : Wherein, the Representing a traffic violation index, c representing the number of vehicles for which traffic violations occur at the last half-year road intersection s n , d representing the total number of vehicles passed by the last half-year road intersection s n ; (6) Model detection Determining according to (1) the likelihood that each road intersection meets the longer waiting time of the red light of the next road intersection : (1) Wherein, the Indicating that the next passing road intersection from the current road intersection satisfies the red light waiting time is long, Representing the max-min operation of the fuzzy matrix, Representing diagonal elements as Is a diagonal matrix of P + , P + is an intermediate variable, D represents a matrix of Column vectors composed of all elements on the main diagonal; determining according to (2) the likelihood that each road intersection meets the condition that no traffic violation will occur : (2) Wherein, the Indicating that all road intersections starting from the current road intersection always meet that no traffic violations will occur, Representation of Is the maximum motionless point function of (2), Z represents the function Is arranged on the upper surface of the base plate, Representing diagonal elements as Is a diagonal matrix of (a); Determining according to equation (3) that each road intersection satisfies the likelihood that the vehicle will eventually pass at a faster speed : (3) Wherein, the V denotes that at some point the vehicle will be travelling at a faster speed, Representing the identity matrix of the cell, Representing diagonal elements as Is a diagonal matrix of (a); (7) Assessing the quality of a road intersection Evaluating the merits of the road intersection s n according to the formula (4) : (4) Where e represents the weight occupied by the average passing speed of the vehicle, Indicating the likelihood that the vehicle will eventually travel at a faster speed, f indicates the weight taken by the average waiting time of the red light, Indicating the likelihood that the next road intersection red light will have a longer waiting time, g indicating the weight taken up by the occurrence of a traffic violation, E 1 、E 3 is a positive indicator for evaluating road intersection s n , and E 2 is a negative indicator for evaluating road intersection s n ; (8) Forecasting optimal paths 1) Determining a set C where a subsequent road intersection is located The set C of the subsequent road intersections of the road intersection s n is as follows: C , wherein s n+q represents the subsequent intersection of the road intersection s n ; 2) Determining an evaluation set D of a subsequent road intersection For each road intersection in set C Calculating its evaluation value according to (4) The evaluation set D of the subsequent road intersection is obtained as follows: D Wherein f (s n+q ) represents the quality evaluation value of the road intersection s n+q ; 3) Determining a next intersection for a vehicle to pass Selecting the evaluation set D with the largest evaluation value Corresponding road intersection As the next road intersection through which the vehicle passes; 4) Outputting the optimal path And by analogy, gradually selecting the next optimal road intersection to obtain a local optimal path.
2. 2. The vehicle path prediction method based on the generalized likelihood Kripke model according to claim 1, wherein the step (2) constructs a path transfer distribution matrix as follows: The path transfer distribution matrix P is constructed as follows: P= Wherein the method comprises the steps of Representing the path-clearance index between road intersections s m and s n , The value is 0.5.

Description

Vehicle path forecasting method based on generalized likelihood Kripke model Technical Field The invention belongs to the technical field of computers, and particularly relates to vehicle path navigation. Background The traditional optimal travel path planning method is mostly a single target optimization method aiming at minimizing the path length, is obtained based on simple iterative computation, but the typical indexes in the urban road are often more than one, and have uncertainty and difference. Therefore, finding the optimal travel path directly by using the data driving mode gradually becomes a research hotspot. The selection of the vehicle position transition may generally describe the actual selection behavior of the driver during travel, and most of the studies based on the vehicle position transition utilize a more complicated machine learning process, which tends to be inefficient for finding the optimal path. In the actual driving process, the selection of the optimal path by the driver can be assumed to be implied in the historical trip data, so that the corresponding driving condition can be acquired and the optimal path planning can be performed based on the historical trip condition under the specific condition. Probability model detection mainly deals with the model detection problem of uncertainty systems generated by stochastic processes, the goal of which is to determine the accuracy of the probability system for quantitative probability specifications. Multi-value model detection mainly deals with model detection problems involving incomplete or inconsistent information systems. Fuzzy model detection mainly deals with model detection problems involving data representation uncertainty systems, which are more focused on truth values of the system on attributes. While the likelihood computation tree logic is more expressive than computation tree logic, it is too restrictive. Some uncertainties may be described using likelihood theory, but cannot be handled directly using likelihood computation tree logic model detection, such as those by the likelihood Kripke structural modeling system with fuzzy tag values. Considering that most of drivers can select the optimal path for traveling, on the premise that all drivers consider that the drivers select the optimal path, the optimal path planning is carried out according to the historical traveling data for traveling between a specific starting point and a specific terminal point. Therefore, the optimal path prediction method which is detected through the model and is based on the generalized likelihood Kripke model can consider the influence of different traffic conditions, and takes the path of the maximum probability evaluation value as the prediction result of the optimal path. In the technical field of highway traffic, one technical problem to be solved urgently at present is to provide a smooth and safe path forecasting method for traveling vehicles. Disclosure of Invention The technical problem to be solved by the invention is to overcome the defects of the prior art, and provide the vehicle path forecasting method based on the generalized likelihood Kripke model, which has the advantages of easy data acquisition, simple and convenient execution process and accurate prediction. The technical scheme adopted for solving the technical problems is as follows: (1) Constructing an actual road condition model Collecting road condition data between a starting point and a destination, and collecting all passing road intersections S between the starting point and the destination: S= Wherein, the And (3) representing any element in the set S, and n represents the number of road intersections. (2) Constructing a path transfer distribution matrix The path transfer distribution matrix P is constructed as follows: P= Wherein s m and s n represent road intersections, Representing the path-clearance index between road intersections s m and s n,。 (3) Quantifying initial traffic conditions The initial traffic condition at road intersection s n is quantized to column vector I n as follows: , Wherein i n is 1, and the remaining elements are 0, which means that the first road intersection through which the vehicle passes is s n. (4) Construction of atomic propositions for vehicle path forecast The vehicle path planning atomic proposition comprises a vehicle average speed v, a traffic light average waiting time t and a traffic violation index p. (5) Quantizing the tag value Quantifying vehicle average passing speed label value as follows: Wherein, the Representing the average passing speed of the vehicle, v max representing the maximum speed of all vehicles passing through the road intersection s n in the last half year, v min representing the minimum speed of all vehicles passing through the road intersection s n in the last half year, and the average waiting time tag value of the red light is quantized as follows: Wherein t represents the average waiting ti