Optimal path planning method in intelligent traffic system

Abstract

The invention relates to the technical field of path planning, discloses an optimal path planning method in an intelligent traffic system, and aims to solve the defects in the aspects of dynamic response, target optimization, multi-scene adaptation and the like in the prior art. According to the technical key points, real-time traffic data, historical traffic data and environmental data are collected, multi-source information is fused, a multi-target optimization model is constructed, and factors such as passing time, cost and carbon emission are comprehensively considered. The system adopts a reinforcement learning algorithm and a dynamic traffic prediction model, and combines a real-time feedback mechanism to realize dynamic update of a path. According to the method, path selection can be optimized in different scenes, such as blockage avoidance in peak hours, flow regulation and control in festivals and holidays, logistics distribution path planning, school bus scheduling, new energy vehicle path optimization and the like. Compared with a traditional method, the method can significantly improve the accuracy, real-time performance and resource utilization efficiency of path planning, has good social and economic benefits, is suitable for various complex traffic scenes, and has wide popularization and application prospects.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and in particular to an optimal path planning method in an intelligent transportation system. Background Art

[0002] With the rapid advancement of urbanization and the continuous growth of the number of motor vehicles, traffic congestion has become one of the major challenges in urban management worldwide. Traffic problems not only reduce the efficiency of urban operations, but also lead to energy waste, environmental pollution and a decline in the quality of life of residents. Therefore, how to alleviate traffic pressure through efficient path planning technology is an important research direction in the field of Intelligent Transportation System (ITS).

[0003] At present, traditional path planning methods mainly rely on classic algorithms (such as Dijkstra algorithm and A* algorithm) to calculate the shortest path or the shortest time path. These methods perform well in static traffic environments, but have the following significant shortcomings in practical applications:

[0004] Traditional methods rely on fixed road network structures and historical traffic data, and are difficult to process real-time traffic information. For example, congested sections during peak hours, traffic interruptions caused by temporary accidents, and changes in road capacity under severe weather conditions may all affect the actual traffic efficiency of the path. However, traditional path planning methods cannot update path calculation results in real time, nor can they respond quickly in combination with complex dynamic traffic data.

[0005] Users have various needs for route planning, including shortest time, lowest cost, least fuel consumption, or lowest carbon emissions. However, existing route planning algorithms usually only focus on a single goal (such as shortest distance or time), ignoring the personalized needs of users and the balance between multiple goals. For example, for logistics and transportation companies, reducing fuel costs and reducing carbon emissions may be priority goals, while for ordinary commuters, time costs may be more important. Traditional methods cannot optimize multiple goals at the same time, nor can they dynamically adjust optimization strategies based on user preferences.

[0006] Existing path planning methods are often based on the calculation of the local optimal path from one point to another, ignoring the global optimization of the entire traffic network. For example, when multiple users simultaneously choose the shortest time path, some sections of the road may be severely congested, while other road resources are not fully utilized. This situation further exacerbates the unbalanced use of urban traffic resources and fails to achieve an improvement in overall traffic efficiency.

[0007] Some path planning systems try to combine historical data for trend prediction, but most methods only stay at the level of simple static analysis and linear regression. In fact, traffic flow has significant nonlinear characteristics and is affected by a variety of external factors (such as weather, major events, and holidays). Traditional methods are weak in dealing with these complex factors, resulting in insufficient prediction accuracy of path planning results and failure to provide users with high-quality navigation services.

[0008] Most current path planning technologies rely on a single algorithm (such as the classic graph search algorithm) and have poor adaptability to complex scenarios. For example, when it comes to multi-objective optimization, dynamic traffic changes, and personalized user needs, a single algorithm is difficult to meet actual needs. In addition, these algorithms are computationally inefficient when dealing with large-scale urban traffic networks and are difficult to meet the requirements of real-time path planning.

[0009] Therefore, we urgently need to design the optimal path planning method in the intelligent transportation system to solve the above problems. Summary of the invention

[0010] The purpose of the present invention is to provide an optimal path planning method in an intelligent transportation system in view of the deficiencies of the prior art, so as to solve the problems raised in the background technology.

[0011] To achieve the above object, the present invention provides the following technical solution: an optimal path planning method in an intelligent transportation system, comprising the following steps:

[0012] Collect traffic data, including real-time traffic data R(t), historical traffic data H(x) and environmental data E(t);

[0013] Construct a multi-objective optimization function based on time, cost and carbon emissions:

[0014] F(x)=w1T(x)+w2C(x)+w3E(x)

[0015] Among them, T(x) is the travel time, C(x) is the toll, E(x) is the carbon emission, and w1, w2, and w3 are weight coefficients;

[0016] Use reinforcement learning algorithm to optimize path planning through state transfer function:

[0017] S t+1 =π(S t , A t )+∈

[0018] Among them, S t is the current state, A t is the current action strategy, π is the strategy function, ∈ is the random disturbance term;

[0019] Dynamically update the path planning results in real time according to the user's personalized needs, and adjust the weights and states in the optimization model based on data feedback.

[0020] As a preferred technical solution of the present invention, the traffic data collection module collects data through traffic cameras, IoT devices and user mobile terminals, and performs data cleaning and standardization processing on R(t), H(x) and E(t), which is defined as:

[0021] D=α1R(t)+α2H(x)+α3E(t)

[0022] Among them, α1, α2, and α3 are standardized coefficients.

[0023] As a preferred technical solution of the present invention, the travel time T(x) of the multi-objective optimization function is calculated by the following formula:

[0024]

[0025] Among them, d i is the distance of the i-th segment, v i is the average vehicle speed of the i-th section, and n is the total number of sections.

[0026] As a preferred technical solution of the present invention, the toll C(x) in the multi-objective optimization function is calculated by the following formula:

[0027]

[0028] Among them, c i，toll is the toll for the i-th segment, c i，fuel is the fuel cost for the i-th segment.

[0029] As a preferred technical solution of the present invention, the carbon emissions E(x) in the multi-objective optimization function is calculated by the following formula:

[0030]

[0031] Among them, e i is the unit carbon emission of the i-th journey, d i For the journey distance.

[0032] As a preferred technical solution of the present invention, the reward function R in the reinforcement learning algorithm is defined by the following formula:

[0033] R=-(w1T(x)+w2C(x)+w3E(x))

[0034] The negative sign indicates that the optimization objective is minimization.

[0035] As a preferred technical solution of the present invention, the dynamic update of the path is realized by a time series prediction model, in which a time series prediction model based on LSTM is adopted to predict the traffic state P(t+k) in the future k time period:

[0036] P(t+k)=f(R(t),H(x),E(t))

[0037] Where f is a nonlinear prediction function.

[0038] As a preferred technical solution of the present invention, the user's personalized needs are achieved by adjusting the weights w1, w2, w3 in the optimization function, which are defined as follows:

[0039] w1+w2+w3=1,w i ∈[0, 1]

[0040] Users can dynamically adjust by selecting their preferences through the user interface.

[0041] As a preferred technical solution of the present invention, the disturbance term in the state transfer function ∈ Generated via normal distribution:

[0042]

[0043] Among them, σ is the variance that controls randomness.

[0044] As a preferred technical solution of the present invention, the final path planning result is determined by the following formula:

[0045]

[0046] Among them, X* is the optimal path, and the optimization goal is to select the path with the lowest cost among all candidate paths.

[0047] Compared with the prior art, the present invention has the following beneficial effects:

[0048] First, the present invention uses multi-source data fusion technology to organically integrate real-time traffic data, historical traffic data and environmental data, and builds a multi-objective optimization model that can comprehensively consider key factors such as time, cost and carbon emissions. Users can dynamically adjust weights according to their own needs to achieve personalized route recommendations, which improves the flexibility and adaptability of route planning.

[0049] Second, the present invention uses a reinforcement learning algorithm and a dynamic traffic prediction model, combined with a real-time state feedback mechanism, which can quickly respond to dynamic changes in complex traffic environments, such as peak congestion, sudden accidents, and bad weather, and achieve real-time optimization of routes. Compared with traditional methods, the present invention significantly improves the accuracy and real-time performance of route planning, and reduces the time and energy loss caused by traffic congestion.

[0050] Third, the present invention is widely applicable to multi-scenario traffic needs, including avoiding traffic jams during peak hours, optimizing traffic flow during holidays, logistics distribution route planning, school bus dispatching, and new energy vehicle route optimization. By improving the overall efficiency of traffic resource utilization and reducing carbon emissions, the present invention is of great significance to relieving traffic pressure, saving energy, and improving the environment, and at the same time has good social value and economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a flow chart of the optimal path planning method in the intelligent transportation system proposed by the present invention. DETAILED DESCRIPTION

[0052] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0053] The following is combined with Figure 1 With reference to the accompanying drawings and a number of embodiments, specific embodiments of the present invention are described in detail.

[0054] Implementation method 1: basic implementation of traffic navigation system;

[0055] Data collection and processing: Traffic data is collected through IoT devices (such as traffic cameras, vehicle sensors, and user mobile devices), including: Real-time traffic data R(t): including the traffic volume and average speed v of the current road section i and traffic congestion index; historical traffic data H(x): including the average daily travel time of each road section and the traffic flow distribution during historical peak periods; environmental data E(t): including weather conditions (such as rainy weather), road construction information, etc.

[0056] The data is standardized and defined as:

[0057] D=α1R(t)+α2H(x)+α3E(t)

[0058] Among them, α1, α2, and α3 are standardization coefficients, which are dynamically adjusted through data training to balance the weights of different data sources.

[0059] Multi-objective optimization model: Building optimization function:

[0060] F(x)=w1T(x)+w2C(x)+w3E(x)

[0061] Indicates the travel time;

[0062] Indicates the cost;

[0063] Represents carbon emissions.

[0064] Path planning algorithm: The system sets weights w1, w2, w3 according to user preferences, and the weights satisfy w1+w2+w3=1.

[0065] For example, when the user prefers the shortest time, the system sets w1 to 0.7, while w2 and w3 are 0.2 and 0.1 respectively.

[0066] Based on the reinforcement learning algorithm, the path is determined by the following state transfer function:

[0067] S t+1 =π(S t , A t )+∈

[0068] S t : Current road status, including traffic flow and average vehicle speed; A t : Action strategy, i.e. the next road segment to be selected;

[0069] is a normally distributed random perturbation term used to explore the optimal solution in path planning.

[0070] Dynamic path update: Combined with the LSTM time series prediction model, the traffic status of the next k time periods is predicted:

[0071] P(t+k)=f(R(t),H(x),E(t))

[0072] The system updates the path planning results every 5 minutes.

[0073] Navigation result output: The optimal path X* is displayed through the user interface, where:

[0074]

[0075] Users can view the path and make dynamic adjustments in real time on mobile devices.

[0076] Implementation method 2: Application of peak-hour route planning;

[0077] During the morning rush hour in the city (e.g. 7:00-9:00 am), the system focuses on optimizing the travel time T(x). Data collection shows the average vehicle speed v during the peak period i Reduced to 30km / h, the system calculates the travel time based on the following formula:

[0078]

[0079] Assume that a path has 5 segments d1, d2, ..., d5, where v i The speeds are 30, 25, 20, 30 and 15 km / h respectively, and the total distance d is 20 km. The travel time is calculated as:

[0080]

[0081] The system prioritizes routes with shorter travel times and dynamically adjusts routes to avoid traffic congestion.

[0082] Implementation method 3: long-distance freight energy-saving route optimization;

[0083] In the long-distance freight application scenario, users prefer low-cost and low-carbon emission route planning, and the system sets the weights to w1=0.2, w2=0.4, and w3=0.4.

[0084] Cost calculation: Assume that a route consists of 3 sections, and the toll for each section is C i,toll and fuel costsc i,fuel They are: c 1,toll =10, c 1,fuel =20; c 2,toll =15, c 2,fuel =30; c 3,toll =20, c 3,fuel =40

[0085] The total cost is:

[0086] C(x)=(10+20)+(15+30)+(20+40)=135

[0087] Carbon emission calculation: Assume that the unit carbon emission of each journey is e i is 0.2kg / km, and the distances d1, d2, and d3 are 50km, 100km, and 150km respectively, then the carbon emissions are:

[0088] E(x)=0.2·50+0.2·100+0.2·150=60kg

[0089] Result output: The system calculates the optimization target F(x), and selects the path with the lowest cost and carbon emissions based on dynamic traffic forecasts, and provides optimization suggestions to users.

[0090] Implementation method 4: Application of personalized route recommendation;

[0091] The user selects the "time priority" mode through the system interface, and the system automatically adjusts the weights w1=0.8, w2=0.1, and w3=0.1.

[0092] Based on real-time traffic data, the system recommends a fast route that avoids congested areas, and adjusts weights through a reinforcement learning model to dynamically update the route to ensure the shortest possible time.

[0093] Implementation method 5: route optimization based on weather conditions;

[0094] This embodiment describes an optimization method for route planning under complex weather conditions, focusing on the impact of rainy and snowy weather on travel time, cost, and carbon emissions.

[0095] Data collection and processing: The system collects weather data W(t) through IoT devices and weather forecast services, including precipitation, road slipperiness index, visibility, etc. The vehicle speed prediction value is adjusted by the following formula:

[0096] v i ′=v i (1-βW(t))

[0097] Among them, v′ i is the adjusted vehicle speed, and β is the weather influence coefficient. The system dynamically calculates the travel time T(x) of different road sections according to the actual weather conditions.

[0098] Path planning model adjustment: In heavy rain, users usually pay more attention to safety and travel time. The system increases the safety weight w4 and adjusts the optimization function to:

[0099] F′(x)=w1T(x)+w2C(x)+w3E(x)+w4S(x)

[0100] Among them, S(x) represents the safety index, which is calculated by the road slipperiness index and historical accident rate.

[0101] Route recommendation results: By adjusting the optimization weights, the system avoids slippery roads with high accident rates and gives priority to recommending safer routes. For example, when the precipitation reaches 50 mm, the system will reduce the priority of viaducts and roads without drainage facilities through dynamic weight allocation, and recommend routes with higher terrain and lower traffic flow.

[0102] Implementation method 6: Path optimization in logistics and distribution scenarios;

[0103] This embodiment is based on logistics distribution, optimizes costs and carbon emissions, and is suitable for long-distance and multi-site distribution.

[0104] Multi-point path planning: The system aims to solve the multi-site problem in logistics distribution by combining an improved reinforcement learning algorithm with a dynamic prediction model to achieve optimal path planning.

[0105] The initial path is determined by the following total cost calculation:

[0106] F(x)=w2C(x)+w3E(x)

[0107] C(x) includes fuel costs and tolls; E(x) is the carbon emissions of the vehicle.

[0108] Dynamic route adjustment: During the delivery process, the route is adjusted according to real-time traffic conditions. For example, when a road section is found to be congested, the system recalculates the optimal route in real time and dynamically pushes the updated delivery route through the vehicle terminal.

[0109] Actual application case: A logistics company needs to deliver goods from a warehouse to five distribution points, with a total journey of 300 kilometers. The initial planned route shows: the total fuel cost is 300 yuan; the carbon emissions are 60kg. When planning the route, the system predicts congestion on the main road and recommends an alternative route, ultimately saving 30 yuan in fuel costs and reducing carbon emissions by 5kg.

[0110] Implementation method 7: Path optimization for new energy vehicles;

[0111] With the popularity of new energy vehicles (such as electric vehicles), their path planning needs to consider the location of charging stations and battery life. This embodiment optimizes the path according to the characteristics of new energy vehicles.

[0112] Data collection and processing: The system collects the remaining power E of the vehicle r and battery range R m , combined with the charging station distribution data C in the map s , construct the following constraints:

[0113] R m ≥d(x) or E r ≥E c (x)

[0114] Among them, d(x) is the total distance of the road section, E c (x) is the amount of electricity required for the path.

[0115] Path planning and optimization: The system gives priority to paths with densely distributed charging stations along the way, while adding the cost of charging station stops in the optimization model:

[0116] F(x)=w1T(x)+w2C(x)+w3E(x)+w5P(x)

[0117] Among them, P(x) represents the time cost of stopping at the charging station.

[0118] Actual application case: An electric vehicle needs to travel from A to B, with a remaining range of 100 km and a target route of 120 km. The system recommends a route that passes through a charging station and calculates the charging time to be 15 minutes, which increases the total travel time by 20 minutes, but avoids the risk of stopping due to insufficient power.

[0119] Implementation method 8: Route planning based on holiday traffic characteristics;

[0120] This embodiment illustrates how to optimize route planning during holidays (such as the Spring Festival and National Day holidays), and is suitable for long-distance self-driving and tourism scenarios.

[0121] Data collection and prediction: The system combines historical holiday data f (x) and real-time traffic data R(t), use the LSTM model to predict holiday traffic flow:

[0122] P f (t+k)=f(H f (x), R(t))

[0123] Among them, P f (t+k) is the traffic flow prediction result for the future k time period.

[0124] Optimization model adjustment: In view of the free highway policy during holidays, the system reduces the toll weight w2;

[0125] The optimization model prefers the time-first path:

[0126] F(x)=w1T(x)+w3E(x)

[0127] Actual application case: During the National Day holiday, a user plans to drive from Beijing to Shanghai. The system predicts that the Beijing-Shanghai Expressway may be congested and recommends avoiding the Beijing-Taipei Expressway in the high-traffic area. Although the total mileage increases by 50 kilometers, the travel time is reduced by 1 hour.

[0128] Implementation method 9: Route planning for school bus dispatch in education districts;

[0129] This example illustrates how to improve transportation efficiency by optimizing the route in school bus scheduling in a specific educational area. d , school bus capacity V c and school start time constraints T s .

[0130] The optimization model considers the following constraints:

[0131] T(x)≤T s ,S d ≤V c

[0132] Path Planning and Scheduling Optimization Through the reinforcement learning model, the system simultaneously optimizes the school bus path and shift schedule, with the goal of minimizing travel time and driving distance.

[0133] Practical application case: There are 3 school buses and 50 students in an educational district. The system assigns students to different vehicle routes through path planning and ensures that all students arrive at school within 30 minutes.

[0134] Implementation method 10: Corporate commuting plan based on carbon emission constraints;

[0135] This embodiment aims at the commuting needs of enterprises and reduces the carbon emissions of employees commuting to and from get off work by optimizing the routes.

[0136] Carbon emission calculation: The system collects vehicle data of enterprise employeesV e , average fuel consumption F e and daily commuting distance D c , calculate the total carbon emissions of employees:

[0137]

[0138] Optimization plan recommendation: Prioritize public transportation and carpooling routes; optimize commuting time based on company parking capacity restrictions.

[0139] Actual application case: The daily commuting carbon emissions of 300 employees of a company were 200kg. After the system recommended an optimization plan, the carbon emissions were reduced to 150kg, and the average commuting time was reduced by 10 minutes.

[0140] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in the field. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An optimal path planning method in an intelligent transportation system, characterized in that: The following steps are involved: Collect traffic data, including real-time traffic data R(t), historical traffic data H(x) and environmental data E(t); Construct a multi-objective optimization function based on time, cost and carbon emissions: F(x)=w1T(x)+w2C(x)+w3E(x) Among them, T(x) is the travel time, C(x) is the toll, E(x) is the carbon emission, and w1, w2, and w3 are weight coefficients; Use reinforcement learning algorithm to optimize path planning through state transfer function: S t+1 =π(S t ,A t )+∈ Among them, S t is the current state, A t is the current action strategy, π is the strategy function, ∈ is the random disturbance term; Dynamically update the path planning results in real time according to the user's personalized needs, and adjust the weights and states in the optimization model based on data feedback.

2. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The traffic data collection module collects data through traffic cameras, IoT devices and user mobile terminals, and performs data cleaning and standardization on R(t), H(x) and E(t), which is defined as: D=α1R(t)+α2H(x)+α3E(t) Among them, α1, α2, and α3 are standardized coefficients.

3. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The travel time T(x) of the multi-objective optimization function is calculated by the following formula: Among them, d i is the distance of the i-th segment, v i is the average vehicle speed of the i-th section, and n is the total number of sections.

4. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The toll C(x) in the multi-objective optimization function is calculated by the following formula: Among them, c i,toll is the toll for the i-th segment, c i,fuel is the fuel cost for the i-th segment.

5. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The carbon emissions E(x) in the multi-objective optimization function are calculated by the following formula: Among them, e i is the unit carbon emission of the i-th journey, d i For the journey distance.

6. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The reward function R in the reinforcement learning algorithm is defined by the following formula: R=-(w1T(x)+w2C(x)+w3E(x)) The negative sign indicates that the optimization objective is minimization.

7. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The dynamic update of the path is realized through the time series prediction model, in which the LSTM-based time series prediction model is used to predict the traffic status P(t+k) in the future k time period: P(t+k)=f(R(t),H(x),E(t)) Where f is a nonlinear prediction function.

8. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: User personalized needs are achieved by adjusting the weights w1, w2, w3 in the optimization function, which are defined as follows: <h2 style=";text-align:left;direction:ltr">w1+w2+w3=1,w<h2 style=";text-align:left;direction:ltr"> i <h2 style=";text-align:left;direction:ltr"> ∈[0,1] Users can dynamically adjust by selecting their preferences through the user interface.

9. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The disturbance term ∈ in the state transfer function is generated by normal distribution: Among them, σ is the variance that controls randomness.

10. The optimal path planning method in an intelligent transportation system according to claim 1, characterized in that: The final path planning result is determined by the following formula: Among them, X * is the optimal path, and the optimization goal is to select the path with the lowest cost among all candidate paths.

Images