Intelligent driving multi-target path planning method based on fuzzy game optimization

Abstract

The invention discloses an intelligent driving multi-target path planning method based on fuzzy game optimization, and the method comprises the following steps: S1, collecting the real-time environment data of a vehicle, and carrying out the preprocessing; s2, constructing a current road environment model, obtaining candidate paths, and quantifying an optimization target of each candidate path; s3, performing fuzzification processing on the optimization target through fuzzy logic to obtain an optimization target weight of each candidate path; s4, constructing a game model, and calculating a game optimization value for each candidate path; s5, performing multi-objective optimization by adopting an improved particle swarm optimization algorithm to generate an optimal path; and S6, executing the optimal path through a vehicle control system, and performing navigation guidance. According to the method, the fuzzy logic, game optimization and the improved particle swarm optimization algorithm are combined, multi-target path planning is achieved, and the method has the advantages of being efficient, safe and intelligent.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving, and particularly to a multi-objective path planning method for intelligent driving optimized based on fuzzy game theory. Background Art

[0002] With the rapid development of intelligent driving technology, autonomous driving has gradually become an important part of future transportation. In an intelligent driving system, path planning, as one of the core tasks, directly affects the driving efficiency, safety, and user experience of the vehicle. The goal of path planning is not only to select the shortest path from the starting point to the ending point, but also to comprehensively consider factors such as road conditions, traffic flow, traffic signals, obstacles, weather conditions, and the behaviors of surrounding traffic participants. Traditional path planning methods usually rely on single-objective optimization and ignore the trade-off between different optimization objectives and the complexity of multi-party games. Therefore, how to achieve efficient, safe, and comfortable path planning in a dynamic and complex traffic environment has become a key problem urgently to be solved in the current intelligent driving technology.

[0003] Existing path planning methods can be roughly divided into three types: graph-based algorithms, optimization algorithms, and artificial intelligence-based algorithms. Graph-based path planning methods, such as Dijkstra's algorithm and A* algorithm, search for the shortest path by constructing a traffic network graph and are simple and effective, widely used in traditional path planning systems. However, these algorithms often cannot adapt to dynamic traffic environments. Although some improved versions of the algorithms can consider traffic flow and road obstacles, they still lack sufficient consideration for complex factors such as multi-objective optimization, driver preferences, and environmental adaptability, and the computational efficiency is low in large-scale urban traffic networks.

[0004] Optimization algorithm-based path planning methods introduce the concept of multi-objective optimization, aiming to optimize multiple objectives simultaneously, such as the shortest time, the lowest energy consumption, the highest safety, etc. Common optimization algorithms include genetic algorithms, particle swarm optimization (PSO), simulated annealing algorithms, etc. These methods have certain advantages in dealing with multi-objective optimization and can balance the conflicts between different objectives by setting the weight coefficients of the objective functions. However, such methods usually rely on predefined weight settings and lack flexible adaptability in the face of dynamic traffic environments. Therefore, how to dynamically adjust the weights and respond to environmental changes in real time remains a bottleneck for optimization algorithms in practical applications.

[0005] In recent years, artificial intelligence-based path planning methods have gradually emerged, especially the introduction of technologies such as deep learning and reinforcement learning, which enable path planning systems to autonomously learn and make decisions based on historical data and real-time feedback. These methods can adaptively adjust path selection strategies through a large amount of data training and continuously optimize decisions in a changing environment. However, artificial intelligence algorithms rely on a large amount of high-quality training data in practical applications, and the training and debugging process of the model is relatively complicated. At the same time, how to combine path planning with factors such as driver preferences, traffic participant behavior, and traffic regulations to make path planning more intelligent and personalized is also a difficulty in current technology.

[0006] However, most of the existing path planning methods rely on static or single-objective optimization, and fail to fully consider the multi-objective optimization needs in dynamic environments. Especially in complex traffic environments, how to comprehensively consider multiple goals, such as safety, time efficiency, energy efficiency, and comfort, and be able to flexibly adjust according to factors such as the behavior of traffic participants, real-time traffic conditions, and weather changes, is still a technical problem that has not been fully solved.

[0007] In addition, the application of game theory in existing technologies is relatively limited. Although game theory has been widely used in fields such as economics and sociology, its application in the field of intelligent driving, especially in multi-party game models, is still in its infancy. Game models are usually used to simulate the interaction and decision-making between different participants. Especially in multi-party games, the behaviors and decisions of participants influence each other. Therefore, how to accurately model the behavior of traffic participants and optimize different paths in combination with game theory is a difficulty that needs to be solved in existing technologies. Especially in a dynamic environment, the behavior patterns and decisions of traffic participants may change at any time. How to adjust the path planning strategy in real time through the game model to ensure the optimality and safety of the path is still a difficulty in current technology.

[0008] Existing path planning methods often ignore the combined application of fuzzy logic and game theory. In complex traffic environments, many factors are fuzzy and uncertain, and traditional mathematical models have difficulty dealing with this uncertainty. Fuzzy logic provides an effective way to deal with uncertainty and ambiguity. It can dynamically adjust the weights of optimization objectives through fuzzy reasoning, and then balance the priorities of different objectives in multi-objective optimization. However, the combined application of fuzzy logic and game models in current technology is still immature, and there is a lack of a comprehensive path planning system that can adjust the weights of optimization objectives in real time, consider the behavior of traffic participants, and perform game optimization.

[0009] Therefore, how to provide an intelligent driving multi-objective path planning method based on fuzzy game optimization is a problem that technical personnel in this field urgently need to solve. Summary of the invention

[0010] An object of the present invention is to propose an intelligent driving multi-objective path planning method based on fuzzy game optimization. The present invention combines fuzzy logic and game optimization theory, uses an improved particle swarm optimization algorithm for multi-objective optimization, dynamically adjusts the weights of optimization objectives, responds to complex traffic environment changes in real time, and solves the multi-party game equilibrium by considering the behaviors and decisions of traffic participants, optimizing path selection. By introducing an adaptive inertia weight and a local search mechanism into the improved particle swarm optimization algorithm, the search efficiency and path planning accuracy are improved, and it has the advantages of high efficiency, adaptability, safety and intelligence.

[0011] An intelligent driving multi-objective path planning method based on fuzzy game optimization according to an embodiment of the present invention includes the following steps:

[0012] S1. Collect real-time environment data of the vehicle and perform preprocessing to generate preprocessed data;

[0013] S2. According to the real-time environment data, construct a current road environment model, obtain candidate paths, and quantify the optimization objectives of each candidate path;

[0014] S3. Perform fuzzy processing on the optimization objectives through fuzzy logic, convert each optimization objective into a fuzzy set, define the membership function of each optimization objective, update the membership function of each optimization objective based on the real-time environment data, dynamically adjust the weights of the optimization objectives, and use fuzzy inference rules to adjust the priorities of each optimization objective to obtain the optimization objective weights of each candidate path in different situations;

[0015] S4. Based on the behaviors and decisions of traffic participants, construct a game model, calculate the behaviors taken by traffic participants and the impact on the selection of candidate paths for each candidate path, and solve the game equilibrium of each candidate path in the multi-party game by combining the game model to obtain the game optimization value of each candidate path after considering the behaviors of traffic participants;

[0016] S5. Combine the optimized objective weights and game optimization values processed by fuzzy logic, calculate the comprehensive evaluation values of each candidate path, use an improved particle swarm optimization algorithm to perform multi-objective optimization on multiple candidate paths, introduce an adaptive inertia weight to dynamically adjust the search ability of particles, and optimize the path in combination with a local search mechanism to generate an optimal path;

[0017] S6. Execute the optimal path through the vehicle control system for navigation guidance, continuously monitor the real-time traffic environment, update the driving strategy in real time through the vehicle control system, adjust the navigation according to the path planning, and reselect the path when encountering sudden changes.

[0018] Optionally, S2 specifically includes:

[0019] S21. Based on the real-time environmental data, identify and classify the current road environmental information. The road environmental information includes traffic flow, traffic obstacles, road surface conditions, the positions and behaviors of traffic participants, and weather conditions. Combine the dynamic changes in traffic flow to construct a current road environmental model. The road environmental model M is represented by a multi-dimensional matrix:

[0020] M = {m ij | i, j ∈ 1, 2,...};

[0021] where M represents the road environmental model, i represents the time period, j represents the road section, and m ij represents the real-time environmental data;

[0022] S22. According to the real-time environmental data, construct the current road network G = (V, E), where V represents the set of nodes in the road network, E represents the set of edges in the road network, and the edge e ij represents the road section connecting the nodes v i and v j . For each edge e ij , define a weight w ij , and the weight is calculated based on traffic flow, weather conditions, road obstacles, traffic participant positions and behaviors, and road surface conditions;

[0023] S23. According to the road environmental model M and traffic flow information, obtain candidate paths. Each candidate path P k is represented by the set of all edges in the path:

[0024] P k = {e 1 , e 2 ,..., e n};

[0025] where P k represents the candidate path, e n represents the edge, and is the nth edge in the path P k ;

[0026] For each candidate path P k , quantify the optimization objectives, and the optimization objectives include safety, time efficiency, energy efficiency, and comfort.

[0027] Optionally, S3 specifically includes:

[0028] S31. Perform fuzzy processing on the optimization objectives through fuzzy logic, convert each optimization objective into a fuzzy set, and for each optimization objective, define a membership function μ iThe membership degree as the optimization objective, and the membership function of each optimization objective is represented by weighted average:

[0029]

[0030] Among them, μ i represents the membership function, α ik represents the weighting coefficient, indicating the contribution degree of the k-th scenario to the objective membership degree, f ik (X) represents the membership function, X represents the environmental characteristics related to the optimization objective, and n represents the number of membership degrees;

[0031] By combining influencing factors through the weighted average method, the fuzzy set of each optimization objective is represented as:

[0032] A i ={μ i1 , μ i2 ,..., μ in};

[0033] Among them, A i represents the fuzzy set of the i-th optimization objective, μ ik represents the membership degree of the i-th optimization objective in the k-th scenario, which is the membership degree value in the k-th scenario, and n represents the number of membership degrees;

[0034] S32. Define the membership function of each optimization objective. For the safety objective, define the membership function μ S as a function of the traffic conditions, road obstacles, and potential risks on the path. Assume the traffic flow is T f , the road obstacle is D o , and the risk level is R. Calculate the membership function of the safety objective:

[0035]

[0036] Among them, μ S represents the membership function of the safety objective, exp represents the natural exponential function, γ 1 , γ 2 , and γ 3 represent adjustment parameters to control the influence of traffic flow, road obstacles, and risk level on the safety objective, T f represents the traffic flow, D o represents the road obstacle, and R represents the risk level;

[0037] For the time efficiency objective, define the membership function μ T as a function of the path length L, traffic flow T f and the passing speed V. Calculate the membership function of the time efficiency objective:

[0038]

[0039] Among them, μ T represents the membership function of the time efficiency target, V represents the passing speed, L represents the path length, and T f represents the traffic flow;

[0040] For the energy efficiency target, define the membership function μ E as a function of the path length L, traffic flow T f and fuel consumption F c to calculate the membership function of the energy efficiency target:

[0041]

[0042] Among them, μ E represents the membership function of the energy efficiency target, and α 1 , α 2 and α 3 represent the weighting coefficients, which are the influence degrees of traffic flow, fuel consumption, and path length on the energy efficiency target respectively. L represents the path length, and F c represents the fuel consumption;

[0043] For the comfort target, define the membership function μ C as a function of the road condition D s , vibration frequency V f and traffic density T d to calculate the membership function of the comfort target:

[0044]

[0045] Among them, μ C represents the membership function of the comfort target, T d represents the traffic density, D s represents the road condition, and α represents the adjustment index to control the influence of vibration frequency and traffic density on the comfort target;

[0046] S33. Based on the real-time environmental data, update the membership degree of each optimization target, and use the real-time environmental data m ij and traffic participant behavior data b ij to adjust each membership function:

[0047] μ' i = μ i · exp(γ i · (m ij + b ij ));

[0048] Among them, μ'i Denote the adjusted membership function, exp represents the natural exponential function, and γ i denotes the adjustment coefficient, which is the influence degree of environmental change on the membership, and μ i denotes the original membership value, which is the membership calculated under the initial situation, and m ij denotes the real-time environmental data, and b ij denotes the traffic participant behavior data;

[0049] S34. Adjust the priority of each optimization objective by combining traffic conditions, road conditions, and weather conditions through fuzzy inference rules:

[0050]

[0051] Among them, P i denotes the priority of optimization objective i, and μ' i denotes the adjusted membership function, exp represents the natural exponential function, and δ i denotes the priority adjustment coefficient, and D o denotes the road condition, and W c denotes the weather condition;

[0052] S35. Dynamically adjust the objective weights according to the real-time environmental data under different situations:

[0053]

[0054] Among them, w i denotes the optimization objective weight, and λ i denotes the adjustment parameter, which controls the influence of priority and membership on the weight, and P i denotes the priority of the optimization objective;

[0055] S36. Calculate the optimization weight of each candidate path according to the adjusted weight:

[0056]

[0057] Among them, W k denotes the optimization weight of candidate path P k , which is the comprehensive score of the candidate path under multi-objective optimization, and w i denotes the weight of optimization objective i, and O ki denotes the normalized value of the i-th optimization objective corresponding to candidate path P k ;

[0058] Optionally, the S4 specifically includes:

[0059] S41. Based on the behaviors and decisions of traffic participants, construct a game model, and set the game participant set A = {a1 , a 2 ,..., a m}, where m in the set of game participants represents the total number of traffic participants, and a m represents the m-th traffic participant. The decision space D of each traffic participant i = {d 1 , d 2 ,..., d k} represents all the actions taken. The utility function of the decision is U i (d k ). The utility function represents the utility value obtained by the i-th traffic participant when choosing the decision d k . The utility function of each participant depends on its own decision and the decisions of the remaining participants, and the payoff function M i is defined as the payoff matrix in the game:

[0060]

[0061] where M i (d k , d j ) represents the payoff function of the i-th traffic participant when choosing the decision d k . T f represents the traffic flow, α i , β i and δ i represent adjustment coefficients to control the influence of various factors on the payoff value. L represents the path length, T c represents the travel time of the selected path, D o represents the road condition, R represents the risk level, γ i represents the adjustment factor related to the interaction behavior, p ij represents the interaction coefficient between traffic participants, λ i represents the adjustment factor affecting the decision, and d j represents the decision;

[0062] S42. According to the game model, calculate the strategy selection of each traffic participant and define the utility function of each traffic participant:

[0063]

[0064] where U i (d k ) represents the utility function of the traffic participant, α ij represents the adjustment factor, which is the influence of traffic flow on the decision of each participant. V ij represents the traffic speed, β ij represents the adjustment factor of the obstacle, and D oijIndicates the impact of obstacles on the path, γ ij Indicates the interaction factor, μ ij Indicates the membership value of traffic participants, λ ij Indicates the adjustment factor of path selection priority;

[0065] S43. Solve the game equilibrium according to the utility function of traffic participants. Assume that the strategy choice of each traffic participant is d k Has a non-linear relationship with the impact on utility, and use a non-linear equation to solve the optimal strategy of each participant:

[0066]

[0067] Among them, U i (d k ) represents the utility of the U i -th traffic participant when choosing the decision d k . c represents the number of decisions, f k represents the weight of the j-th feature, m represents the number of features, w j represents the weight of the j-th feature;

[0068] S44. Calculate the game optimization value of each candidate path according to the behavior and decision of each participant. Let G k be the game optimization value of the candidate path P k , which represents the feasibility and priority of the candidate path P k after considering the behavior of traffic participants:

[0069]

[0070] Among them, G k represents the game optimization value of the candidate path P k , α i represents the influence factor of the i-th traffic participant on the selection of the candidate path, U i (p i ) represents the utility value of the i-th traffic participant when choosing the candidate path P k , θ i represents the adjustment coefficient that affects the decision of the participant, γ ij represents the interaction factor, μ ij represents the membership value of traffic participants when choosing a path.

[0071] Optionally, the S5 specifically includes:

[0072] S51. Combine the optimized objective weight after fuzzy logic processing with the game optimization value. Let W i be the weight of the i-th optimization objective, G k be the game optimization value of the candidate path P kThe game optimization value, for each candidate path P k , calculate the comprehensive evaluation value of each path according to the optimized objective weight and game optimization value after fuzzy logic processing:

[0073]

[0074] Among them, C k represents the comprehensive evaluation value, W i represents the weight of the i-th optimized objective, O ki represents the normalized value of the i-th optimized objective corresponding to the path P k , β represents the adjustment coefficient of the game optimization value, G k represents the game optimization value of the candidate path P k , ζ represents the interaction influence coefficient, interaction(P k ) represents the interaction effect between the candidate path P k and traffic participants, which is the mutual influence between path selections;

[0075] S52. Use the improved particle swarm optimization algorithm to perform multi-objective optimization on the candidate paths. Set the search space of the improved particle swarm optimization algorithm as S = {P 1 , P 2 ,..., P m}, where P m in the search space represents the m-th candidate path, m represents the number of paths, the search space of each candidate path is a particle, and the state of each particle P k is represented by position and velocity. Iteratively update each particle through the improved particle swarm optimization algorithm. The update includes an inertia term, a cognitive term, and a social term:

[0076]

[0077] Among them, represents the velocity of the particle P k in the (t + 1)-th iteration, represents the velocity of the particle P k in the t-th iteration, represents the position of the particle P k in the t-th iteration, represents the position of the particle P k in the (t + 1)-th iteration, P best represents the personal best position of the particle P k , G best represents the global best position, c 1 and c 2 represent learning factors, rand 1 and rand 2represents a random number, ω represents the adaptive inertia weight, which controls the velocity update of the particle;

[0078] S53. Introduce the adaptive inertia weight ω to dynamically adjust the exploration ability of the particle. The inertia weight decreases with the number of iterations:

[0079]

[0080] where ω(t) represents the adaptive inertia weight in the t-th iteration, ω max represents the initial inertia weight, ω min represents the minimum inertia weight, T represents the maximum number of iterations, and t represents the current number of iterations;

[0081] S54. Gradually adjust the path selection strategy during the search process. The improved particle swarm optimization algorithm is optimized through global and local searches, and the comprehensive evaluation value C is updated k :

[0082]

[0083] where, represents the position of particle P k in the (t + 1)-th iteration, represents the position of particle P k in the t-th iteration, λ represents the step size factor, P best represents the position of particle P k 's personal best position, μ represents the local search step size factor, which adjusts the search range of the particle near the local optimal solution, represents the comprehensive evaluation value of path P k after the t-th iteration, represents the comprehensive evaluation value of particle P k 's gradient;

[0084] S55. Through the combined action of multiple search processes and local search mechanisms, gradually improve the overall performance of the candidate path and generate the optimal path:

[0085]

[0086] where, P opt represents the optimal path, represents the comprehensive evaluation value of path P k after the T-th iteration, S represents the search space of the improved particle swarm optimization algorithm, and argmax represents the variable value when the function takes the maximum value.

[0087] The beneficial effects of the present invention are:

[0088] First, by processing multiple optimization objectives in path planning through fuzzy logic, such as safety, time efficiency, energy efficiency, and comfort, the present invention can transform these objectives into fuzzy sets and dynamically adjust the weights of each objective. The introduction of fuzzy inference rules enables flexible adjustment of the objective priorities in different traffic scenarios, thereby achieving the optimization of balancing multiple objectives in an uncertain traffic environment. This method improves the adaptability and real-time response ability of the path planning system in a dynamic and complex environment, ensuring that the vehicle selects the optimal path.

[0089] Secondly, the present invention introduces a game optimization model, which considers the behaviors and decisions of traffic participants. By solving the game equilibrium, it can accurately simulate the interactions and decision-making influences among multiple participants, achieving multi-party game optimization in path selection. This enables path planning to not only consider road conditions and traffic flow but also take into account the behaviors and strategies of other traffic participants, improving the feasibility, safety, and priority selection degree of the path.

[0090] Finally, by adopting an improved particle swarm optimization algorithm, the efficiency and accuracy of path planning are further enhanced. By introducing an adaptive inertia weight and a local search mechanism, the improved particle swarm optimization algorithm can dynamically adjust the search ability of particles, avoid premature convergence to local optimal solutions, and accelerate the convergence process. When optimizing multiple objectives, it can achieve a good balance between global and local searches, thereby realizing fast and accurate path planning.

[0091] In summary, the intelligent driving multi-objective path planning method of the present invention combines fuzzy logic, game optimization, and an improved particle swarm optimization algorithm, can optimize the path in real time in a dynamic and complex traffic environment, has high efficiency, safety, and adaptability, and significantly improves the overall performance and user experience of the intelligent driving system. Description of the Drawings

[0092] The drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation to the present invention. In the drawings:

[0093] Figure 1 is a flowchart of an intelligent driving multi-objective path planning method based on fuzzy game optimization proposed by the present invention;

[0094] Figure 2 is a schematic diagram of an improved particle swarm optimization algorithm for an intelligent driving multi-objective path planning method based on fuzzy game optimization proposed by the present invention. Detailed Embodiments

[0095] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are all simplified schematic diagrams, only schematically showing the basic structure of the present invention, so they only show the components related to the present invention.

[0096] Reference Figure 1 and Figure 2 , an intelligent driving multi-objective path planning method based on fuzzy game optimization, comprising the following steps:

[0097] S1. Collect real-time environmental data of the vehicle, and perform preprocessing to generate preprocessed data;

[0098] S2. According to the real-time environmental data, construct a current road environment model, obtain candidate paths, and quantify the optimization objectives of each candidate path;

[0099] S3. Perform fuzzy processing on the optimization objectives through fuzzy logic, convert each optimization objective into a fuzzy set, define the membership function of each optimization objective, update the membership function of each optimization objective based on the real-time environmental data, dynamically adjust the optimization objective weights, and use fuzzy inference rules to adjust the priority of each optimization objective, so as to obtain the optimization objective weights of each candidate path under different situations;

[0100] S4. Based on the behaviors and decisions of traffic participants, construct a game model, for each candidate path, calculate the behaviors taken by traffic participants and the influence on the selection of candidate paths, and combine the game model to solve the game equilibrium of each candidate path in multi-party games, so as to obtain the game optimization value of each candidate path after considering the behaviors of traffic participants;

[0101] S5. Combine the optimization objective weights and game optimization values processed by fuzzy logic, calculate the comprehensive evaluation value of each candidate path, use an improved particle swarm optimization algorithm to perform multi-objective optimization on multiple candidate paths, introduce an adaptive inertia weight to dynamically adjust the search ability of particles, and combine a local search mechanism to optimize the path to generate an optimal path;

[0102] S6. Execute the optimal path through the vehicle control system for navigation guidance, continuously monitor the real-time traffic environment, update the driving strategy in real time through the vehicle control system, adjust the navigation according to the path planning, and re-select the path when encountering sudden changes.

[0103] In this embodiment, the S2 specifically includes:

[0104] S21. Identify and classify the current road environment information based on the real-time environment data. The road environment information includes traffic flow, traffic obstacles, road surface conditions, the positions and behaviors of traffic participants, and weather conditions. Construct a current road environment model by combining the dynamic changes in traffic flow. The road environment model M is represented by a multi-dimensional matrix:

[0105] M = {m ij | i, j ∈ 1, 2,...};

[0106] where M represents the road environment model, i represents the time period, j represents the road section, and m ij represents the real-time environment data;

[0107] S22. Construct the current road network G = (V, E) according to the real-time environment data. Here, V represents the set of nodes in the road network, and E represents the set of edges in the road network. The edge e ij represents the road section connecting nodes v i and v j . For each edge e ij , define a weight w ij , and the weight is calculated based on traffic flow, weather conditions, road obstacles, traffic participant positions and behaviors, and road surface conditions;

[0108] S23. Obtain candidate paths according to the road environment model M and traffic flow information. Each candidate path P k is represented by the set of all edges in the path:

[0109] P k = {e 1 , e 2 ,..., e n};

[0110] where P k represents the candidate path, e n represents the edge, and is the nth edge in the path P k ;

[0111] For each candidate path P k , quantify the optimization objectives, and the optimization objectives include safety, time efficiency, energy efficiency, and comfort.

[0112] In this embodiment, the specific content of S3 includes:

[0113] S31. Perform fuzzy processing on the optimization objectives through fuzzy logic, convert each optimization objective into a fuzzy set, and for each optimization objective, define a membership function μ i as the membership degree of the optimization objective. The membership function of each optimization objective is represented by a weighted average:

[0114]

[0115] Among them, μ i represents the membership function, and α ik represents the weighting coefficient, indicating the contribution degree of the k-th scenario to the membership degree of the target. f ik (X) represents the membership function, X represents the environmental characteristics related to the optimization target, and n represents the number of membership degrees;

[0116] By combining the influencing factors through the weighted average method, the fuzzy set of each optimization target is expressed as:

[0117] A i ={μ i1 , μ i2 ,..., μ in};

[0118] Among them, A i represents the fuzzy set of the i-th optimization target, μ ik represents the membership degree of the i-th optimization target in the k-th scenario, which is the membership degree value in the k-th scenario, and n represents the number of membership degrees;

[0119] S32. Define the membership function of each optimization target. For the safety target, define the membership function μ S as a function of the traffic conditions, road obstacles, and potential risks on the path. Assume the traffic flow is T f , the road obstacle is D o , and the risk level is R. Calculate the membership function of the safety target:

[0120]

[0121] Among them, μ S represents the membership function of the safety target, exp represents the natural exponential function, and γ 1 , γ 2 , and γ 3 represent the adjustment parameters, controlling the influence of traffic flow, road obstacles, and risk level on the safety target. T f represents the traffic flow, D o represents the road obstacle, and R represents the risk level;

[0122] For the time efficiency target, define the membership function μ T as a function of the path length L, traffic flow T f , and passing speed V. Calculate the membership function of the time efficiency target:

[0123]

[0124] Among them, μT The membership function representing the time efficiency target, V represents the passing speed, L represents the path length, and T f represents the traffic flow;

[0125] For the energy efficiency target, define the membership function μ E as a function of the path length L, the traffic flow T f and the fuel consumption F c to calculate the membership function of the energy efficiency target:

[0126]

[0127] where μ E represents the membership function of the energy efficiency target, α 1 , α 2 and α 3 represent the weighting coefficients, which are the influence degrees of the traffic flow, the fuel consumption, and the path length on the energy efficiency target respectively, L represents the path length, and F c represents the fuel consumption;

[0128] For the comfort target, define the membership function μ C as a function of the road condition D s , the vibration frequency V f and the traffic density T d to calculate the membership function of the comfort target:

[0129]

[0130] where μ C represents the membership function of the comfort target, T d represents the traffic density, D s represents the road condition, and α represents the adjustment index, which controls the influence of the vibration frequency and the traffic density on the comfort target;

[0131] S33. Based on the real-time environmental data, update the membership degree of each optimization target, and use the real-time environmental data m ij and the traffic participant behavior data b ij to adjust each membership function:

[0132] μ' i = μ i · exp(γ i · (m ij + b ij ));

[0133] where μ' i represents the adjusted membership function, exp represents the natural exponential function, and γ iDenotes the adjustment coefficient, which is the influence degree of environmental change on membership degree, μ i Denotes the original membership degree value, which is the membership degree calculated under the initial situation, m ij Denotes the real-time environmental data, b ij Denotes the behavior data of traffic participants;

[0134] S34. Combine traffic conditions, road conditions, and weather conditions through fuzzy inference rules to adjust the priority of each optimization objective:

[0135]

[0136] Among them, P i Denotes the priority of optimization objective i, μ' i Denotes the adjusted membership function, exp represents the natural exponential function, δ i Denotes the priority adjustment coefficient, D o Denotes the road condition, W c Denotes the weather condition;

[0137] S35. Dynamically adjust the target weights according to the real-time environmental data in different situations:

[0138]

[0139] Among them, w i Denotes the optimization target weight, λ i Denotes the adjustment parameter, which controls the influence of priority and membership degree on the weight, P i Denotes the priority of the optimization target;

[0140] S36. Calculate the optimized weight of each candidate path according to the adjusted weight:

[0141]

[0142] Among them, W k Denotes the optimized weight of candidate path P k which is the comprehensive score of the candidate path under multi-objective optimization, w i Denotes the weight of optimization objective i, O ki Denotes candidate path P k The normalized value of the corresponding i-th optimization objective.

[0143] In this embodiment, the S4 specifically includes:

[0144] S41. Based on the behaviors and decisions of traffic participants, construct a game model, and set the game participant set A = {a 1 ,a 2 ,...,a m, where \(m\) in the set of game participants represents the total number of traffic participants, and \(a\) m represents the \(m\)th traffic participant. The decision space \(D\) of each traffic participant i = \(\{d\) 1 , \(d\) 2 ,..., \(d\) k}\) represents all the actions taken. The utility function of the decision is \(U\) i (\(d\) k ) represents the utility value obtained by the \(i\)th traffic participant when choosing the decision \(d\) k . The utility function of each participant depends on its own decision and the decisions of the remaining participants, and the payoff function \(M\) i is defined as the payoff matrix in the game:

[0145]

[0146] where \(M\) i (\(d\) k , \(d\) j ) represents the payoff function of the \(i\)th traffic participant when choosing the decision \(d\) k . \(T\) f represents the traffic flow, \(\alpha\) i , \(\beta\) i and \(\delta\) i represent adjustment coefficients to control the influence of various factors on the payoff value. \(L\) represents the path length, \(T\) c represents the travel time of the selected path, \(D\) o represents the road condition, \(R\) represents the risk level, \(\gamma\) i represents the adjustment factor related to the interaction behavior, \(p\) ij represents the interaction coefficient between traffic participants, \(\lambda\) i represents the adjustment factor affecting the decision, and \(d\) j represents the decision;

[0147] S42. According to the game model, calculate the strategy selection of each traffic participant and define the utility function of each traffic participant:

[0148]

[0149] where \(U\) i (\(d\) k ) represents the utility function of the traffic participant. \(\alpha\) ij represents the adjustment factor, which is the influence of the traffic flow on the decision of each participant. \(V\) ij represents the traffic speed, \(\beta\) ij represents the adjustment factor of the obstacle, and \(D\) oij represents the influence of the obstacle on the path. \(\gamma\) ij represents the interaction factor, and \(\mu\) ijRepresents the membership value of traffic participants, λ ij Represents the adjustment factor of path selection priority;

[0150] S43. Solve the game equilibrium according to the utility function of traffic participants. Assume that the strategy selection d of each traffic participant k Has a non-linear relationship with the impact on utility, and use a non-linear equation to solve the optimal strategy of each participant:

[0151]

[0152] Among them, U i (d k ) Represents the utility of the U i th traffic participant when choosing the decision d k , c represents the number of decisions, f k Represents the weight of the jth feature, m represents the number of features, w j Represents the weight of the jth feature;

[0153] S44. Calculate the game optimization value of each candidate path according to the behavior and decision of each participant. Let G k Be the game optimization value of the candidate path P k , representing the feasibility and priority of the candidate path P k After considering the behavior of traffic participants:

[0154]

[0155] Among them, G k Represents the game optimization value of the candidate path P k , α i Represents the influence factor of the ith traffic participant on the selection of the candidate path, U i (p i ) Represents the utility value of the ith traffic participant when choosing the candidate path P k , θ i Represents the adjustment coefficient affecting the decision of the participant, γ ij Represents the interaction factor, μ ij Represents the membership value of traffic participants when choosing a path.

[0156] In this embodiment, the S5 specifically includes:

[0157] S51. Combine the optimized target weight after fuzzy logic processing with the game optimization value. Let W i Be the weight of the ith optimization target, G k Be the game optimization value of the candidate path P k , for each candidate path P k, calculate the comprehensive evaluation value of each path according to the optimized objective weight and game optimization value after fuzzy logic processing:

[0158]

[0159] Among them, C k represents the comprehensive evaluation value, W i represents the weight of the i-th optimized objective, O ki represents the normalized value of the i-th optimized objective corresponding to the path P k , β represents the adjustment coefficient of the game optimization value, G k represents the game optimization value of the candidate path P k , ζ represents the interaction influence coefficient, interaction(P k ) represents the interaction effect between the candidate path P k and traffic participants, which is the mutual influence between path selections;

[0160] S52. Use the improved particle swarm optimization algorithm to perform multi-objective optimization on the candidate paths. Set the search space of the improved particle swarm optimization algorithm as S = {P 1 , P 2 ,..., P m}, where P m in the search space represents the m-th candidate path, m represents the number of paths, the search space of each candidate path is a particle, and the state of each particle P k is represented by position and velocity. Iteratively update each particle through the improved particle swarm optimization algorithm, and the update includes an inertia term, a cognitive term, and a social term:

[0161]

[0162] Among them, represents the velocity of the particle P k in the (t + 1)-th iteration, represents the velocity of the particle P k in the t-th iteration, represents the position of the particle P k in the t-th iteration, represents the position of the particle P k in the (t + 1)-th iteration, P best represents the personal best position of the particle P k , G best represents the global best position, c 1 and c 2 represent learning factors, rand 1 and rand 2 represent random numbers, and ω represents the adaptive inertia weight, which controls the velocity update of the particle;

[0163] S53. Introduce an adaptive inertia weight ω to dynamically adjust the exploration ability of particles, and the inertia weight decreases with the number of iterations:

[0164]

[0165] Among them, ω(t) represents the adaptive inertia weight in the t-th iteration, ω max represents the initial inertia weight, ω min represents the minimum inertia weight, T represents the maximum number of iterations, and t represents the current number of iterations;

[0166] S54. Gradually adjust the path selection strategy during the search process. The improved particle swarm optimization algorithm is optimized through global and local searches, and the comprehensive evaluation value C is updated k :

[0167]

[0168] Among them, represents the position of particle P k in the (t + 1)-th iteration, represents the position of particle P k in the t-th iteration, λ represents the step size factor, P best represents the personal best position of particle P k and μ represents the local search step size factor, which adjusts the search range of the particle near the local optimal solution. represents the comprehensive evaluation value of path P k after the t-th iteration, represents the gradient of the comprehensive evaluation value of particle P k ;

[0169] S55. Through the combined action of multiple search processes and local search mechanisms, gradually improve the overall performance of the candidate path and generate the optimal path:

[0170]

[0171] opt represents the optimal path, represents the path P k and the final comprehensive evaluation value after the T-th iteration, S represents the search space of the improved particle swarm optimization algorithm, and argmax represents the variable value when the function takes the maximum value.

[0172] Example 1:

[0173] To verify the feasibility of the present invention in implementation, the present invention is applied to the intelligent driving path planning of an urban intelligent transportation system. In this scenario, a typical urban road environment is selected for experiments, simulating the behaviors of multiple traffic participants (such as vehicles, pedestrians, bicycles, etc.) in a complex traffic environment, and using the intelligent driving multi-objective path planning method based on fuzzy game optimization of the present invention to perform path selection and optimization. The experimental site is selected as the core business district of a certain city, where the traffic is intensive and the traffic conditions are complex, presenting certain challenges.

[0174] The experimental site is the bustling commercial center area of the city, where the roads are criss-crossed, the traffic flow is large, the interaction between traffic lights, pedestrians, non-motor vehicles and motor vehicles is complex, and the traffic conditions often change. Specifically, the roads in the area include several main roads and feeder roads, with different widths, significant fluctuations in traffic flow at different time periods, and many road obstacles, such as construction areas, traffic accidents, etc. At the same time, weather factors (such as rainy days, haze weather, etc.) will also affect the rationality of path selection.

[0175] In this scenario, an autonomous vehicle is set to start from a commercial building and the destination is a nearby shopping center. This path needs to avoid traffic congestion areas and avoid passing through road sections under construction, while considering maximizing fuel consumption savings and improving driving comfort. To simulate the actual traffic situation, real-time data is collected through sensors, including current traffic flow, road obstacles, weather conditions, and dynamic behavior data of traffic participants.

[0176] In this experiment, the system first obtains information such as traffic flow, road obstacles, and weather conditions through the real-time environment data collection module. These data are obtained in real time through vehicle sensors and traffic monitoring systems and transmitted to the central control system for preprocessing. The preprocessed data will be used to construct a road environment model and generate multiple candidate paths. Each candidate path will be quantified into multiple optimization objectives, specifically including: safety, time efficiency, energy efficiency, and comfort.

[0177] Through the fuzzy logic module, these optimization objectives are fuzzified. According to the real-time traffic conditions and environmental changes, the fuzzy logic module will dynamically adjust the weights of each optimization objective. For example, in the case of traffic congestion, the weight of time efficiency may be increased, while in the case of poor weather conditions (such as rainy days or haze weather), the weight of the safety objective may increase. In this way, the system will flexibly adjust the priorities of each optimization objective according to the actual situation to ensure the rationality of path selection.

[0178] Through the game optimization module, the system simulates the decision-making behaviors of multiple traffic participants. Considering the impacts of the behaviors of other vehicles, pedestrians, non-motor vehicles and other traffic participants on route selection, the system calculates the game equilibrium of each candidate route through a game model, so as to select the most suitable route. In this process, the system not only considers the behaviors of other traffic participants, but also takes into account their impacts on route selection. For example, when there is traffic congestion, the choices of other vehicles may affect the route selection of autonomous vehicles.

[0179] Finally, the system uses an improved particle swarm optimization algorithm to perform multi-objective optimization on the candidate routes, further improving the accuracy of route selection.

[0180] Table 1 Comparison Table of Experimental Data

[0181]

[0182]

[0183] In terms of time efficiency, most of the routes in the method of the present invention (P1 to P4) are between 15 and 20 minutes, while the time efficiency of the routes selected by the traditional methods (P5, P6) is relatively low, being 20.0 and 21.5 minutes respectively. Especially for route P1 (Main Road A → Feeder Road B → Commercial Building), its time efficiency is 15.2 minutes, which is significantly better than the traditional methods, proving that the present invention can more effectively select unobstructed routes, thereby reducing driving time.

[0184] Secondly, in terms of energy efficiency, the fuel consumption of the method of the present invention is generally low. For example, the fuel consumption of route P1 is 0.08 L / km, which is much lower than 0.12 L / km and 0.13 L / km of the traditional methods. This shows that by optimizing route selection, the present invention can effectively reduce the fuel consumption of vehicles and improve energy efficiency. Route P1 has a good balance between fuel consumption and time efficiency, which is particularly important in urban roads, as it can both save fuel consumption and improve traffic efficiency.

[0185] In terms of the safety score, the routes of the method of the present invention generally have high scores in terms of safety. For example, route P3 (Feeder Road B → Main Road A → Commercial Building) obtained a high safety score of 9 points, while the traditional routes are only 6 to 5 points. This shows that the present invention can better consider traffic conditions and potential traffic risks and select safer driving routes. Route P1 also received a relatively high safety rating of 8 points, further verifying the advantages of the present invention in ensuring driving safety.

[0186] In terms of comfort, the method of the present invention also performs excellently. For example, path P3 obtained 8 points in the comfort score, which is significantly better than the traditional paths (P5 is 4 points and P6 is 5 points). This means that when selecting a path, the present invention not only considers time and energy efficiency, but also fully considers the smoothness of vehicle driving and the comfort of the driver, which is crucial for improving the user experience.

[0187] Finally, in terms of the comprehensive evaluation value, the path selection of the method of the present invention is generally more superior. For example, the final comprehensive evaluation value of path P3 is 0.88, which is much higher than 0.72 and 0.68 of the traditional methods. Generally speaking, the multi-objective optimization method of the present invention can comprehensively consider multiple factors such as safety, time efficiency, energy efficiency, and comfort, so as to provide the optimal path selection.

[0188] In summary, from multiple dimensions such as time efficiency, energy efficiency, safety, comfort, and comprehensive evaluation, the path planning method of the present invention shows significant advantages over the traditional methods. This proves that the intelligent driving multi-objective path planning method based on fuzzy game optimization has high adaptability and practicability in a dynamic traffic environment, and can effectively improve driving efficiency and user experience.

[0189] The above is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, makes equivalent substitutions or changes, and all should be covered within the protection scope of the present invention.

Claims

1. A multi-objective path planning method for intelligent driving based on fuzzy game optimization, characterized in that: The steps include: S1, collect real-time environmental data of the vehicle, perform preprocessing, and generate preprocessed data; S2. constructing a current road environment model based on the real-time environment data, obtaining candidate paths, and quantifying the optimization target of each candidate path; S3, fuzzifying the optimization target through fuzzy logic, converting each optimization target into a fuzzy set, defining the membership function of each optimization target, updating the membership function of each optimization target based on real-time environmental data, dynamically adjusting the optimization target weight, and adjusting the priority of each optimization target using fuzzy inference rules to obtain the optimization target weight of each candidate path in different scenarios; S4. Based on the behaviors and decisions of traffic participants, a game model is constructed. For each candidate path, the behaviors taken by traffic participants and their impact on the selection of candidate paths are calculated. The game equilibrium of each candidate path in the multi-party game is solved by combining the game model, and the game optimization value of each candidate path after considering the behaviors of traffic participants is obtained. S5. Combine the optimization target weight and game optimization value after fuzzy logic processing to calculate the comprehensive evaluation value of each candidate path, use the improved particle swarm optimization algorithm to perform multi-objective optimization on multiple candidate paths, introduce adaptive inertia weight to dynamically adjust the search ability of particles, and optimize the path in combination with the local search mechanism to generate the optimal path; S6. Execute the optimal path through the vehicle control system, provide navigation guidance, continuously monitor the real-time traffic environment, update the driving strategy in real time through the vehicle control system, make navigation adjustments based on the path planning, and reselect the path when encountering sudden changes.

2. The intelligent driving multi-objective path planning method based on fuzzy game optimization according to claim 1 is characterized in that: The S2 specifically includes: S21. Based on the real-time environmental data, identify and classify the current road environment information, the road environment information includes traffic flow, traffic obstacles, road conditions, location and behavior of traffic participants, and weather conditions, and construct a current road environment model in combination with the dynamic changes of traffic flow. The road environment model M is represented by a multidimensional matrix: M={m ij |i,j∈1,2,...}; Among them, M represents the road environment model, i represents the time period, j represents the road section, and m ij Represents real-time environmental data; S22. Construct a current road network G=(V, E) according to the real-time environment data, where V represents a node set of the road network, E represents an edge set of the road network, and edge e ij Represents the connection node v i and v j road segment, for each edge e ij , define the weight w ij , the weights are calculated based on traffic flow, weather conditions, road obstacles, location and behavior of traffic participants, and road conditions; S23, according to the road environment model M and traffic flow information, obtain candidate paths, each candidate path P k It is represented by the set of all edges in the path: P k ={e1,e2,...,e n }; Among them, P k represents the candidate path, e n represents the edge, which is the path P k The nth edge in ; For each candidate path P k , quantify the optimization objectives, which include safety, time efficiency, energy efficiency and comfort.

3. The intelligent driving multi-objective path planning method based on fuzzy game optimization according to claim 1 is characterized in that: The S3 specifically includes: S31, fuzzifying the optimization target by fuzzy logic, converting each optimization target into a fuzzy set, and defining a membership function μ for each optimization target. i As the membership of the optimization target, the membership function of each optimization target is expressed by weighted average: Among them, μ i represents the membership function, α ik represents the weighting coefficient, which indicates the contribution of the k-th situation to the target membership, f ik (X) represents the membership function, X represents the environmental characteristics related to the optimization objective, and n represents the number of memberships; By combining the influencing factors through the weighted average method, the fuzzy set of each optimization objective is expressed as: A i ={μ i1 ,m i2 ,...,m in }; Among them, A i represents the fuzzy set of the i-th optimization objective, μ ik represents the membership of the i-th optimization target in the k-th situation, is the membership value in the k-th situation, and n represents the number of memberships; S32. Define the membership function of each optimization target. For the safety target, define the membership function μ S As a function of the traffic conditions, road obstacles and potential risks on the path, the traffic flow is assumed to be T f , the road obstacle is D o , the risk level is R, and the membership function of the safety target is calculated: Among them, μ S represents the membership function of the safety target, exp represents the natural exponential function, γ1, γ2 and γ3 represent the adjustment parameters to control the impact of traffic flow, road obstacles and risk level on the safety target, T f represents the traffic flow, D o represents road obstacles, and R represents the risk level; For the time efficiency goal, define the membership function μ T As path length L, traffic flow T f and the function of the traffic speed V, calculate the membership function of the time efficiency target: Among them, μ T represents the membership function of the time efficiency target, V represents the travel speed, L represents the path length, T f Indicates traffic flow; For the energy efficiency target, define the membership function μ E As path length L, traffic flow T f and fuel consumption F c Function, calculate the membership function of the energy efficiency target: Among them, μ E represents the membership function of the energy efficiency target, α1, α2 and α3 represent weighting coefficients, which are the influence of traffic flow, fuel consumption and path length on the energy efficiency target, L represents the path length, F c Indicates fuel consumption; For the comfort objective, define the membership function μ C As road condition D s , vibration frequency V f and traffic density T d Function, calculate the membership function of the comfort target: Among them, μ C represents the membership function of the comfort objective, T d Denotes traffic density, D s represents the road condition, α represents the adjustment index, and controls the impact of vibration frequency and traffic density on the comfort target; S33, based on the real-time environmental data, update the membership of each optimization target, and use the real-time environmental data m ij and traffic participant behavior data b ij , adjust each membership function: Among them, μ' i represents the adjusted membership function, exp represents the natural exponential function, γ i Represents the adjustment coefficient, which is the degree of influence of environmental changes on the membership degree, μ i Represents the original membership value, which is the membership calculated under the initial situation, m ij represents real-time environmental data, b ij Represents traffic participant behavior data; S34. By combining traffic conditions, road conditions, and weather conditions through fuzzy reasoning rules, the priority of each optimization target is adjusted: Among them, P i represents the priority of optimization target i, μ' i represents the adjusted membership function, exp represents the natural exponential function, δ i Denotes the priority adjustment coefficient, D o Indicates road conditions, W c Indicates weather conditions; S35. Dynamically adjust target weights based on real-time environmental data in different scenarios: Among them, w i represents the optimization target weight, λ i Represents the adjustment parameter, controlling the influence of priority and membership on weight, P i Indicates the priority of the optimization target; S36. Calculate the optimization weight of each candidate path according to the adjusted weight: Among them, W k Represents the candidate path P k The optimization weight is the comprehensive score of the candidate path under multi-objective optimization, w i represents the weight of optimization target i, O ki Represents the candidate path P k The normalized value of the corresponding i-th optimization objective.

4. The intelligent driving multi-objective path planning method based on fuzzy game optimization according to claim 1 is characterized in that: The S4 specifically includes: S41. Based on the behaviors and decisions of traffic participants, a game model is constructed, and the game participant set A = {a1, a2, ..., a m }, m in the game participant set represents the total number of traffic participants, a m represents the mth traffic participant, and the decision space D of each traffic participant i ={d1,d2,...,d k } represents all actions taken, and the utility function of the decision is U i (d k ), the utility function represents the utility of the i-th traffic participant in choosing decision d k The utility value obtained when the transaction is completed, the utility function of each participant depends on its own decision and the decisions of the remaining participants, and the payment function M is defined i As the payoff matrix in the game: Among them, M i (d k ,d j ) indicates that the i-th traffic participant chooses decision d k The payment function when T f represents the traffic flow, α i , β i and δ i represents the adjustment coefficient, which controls the impact of various factors on the payment value, L represents the path length, T c represents the travel time of the selected path, D o represents the road condition, R represents the risk level, γ i represents the regulatory factor related to the interaction behavior, p ij represents the interaction coefficient between traffic participants, λ i represents the adjustment factor affecting the decision, d j Indicates decision making; S42. According to the game model, calculate the strategy choice of each traffic participant and define the utility function of each traffic participant: Among them, U i (d k ) represents the utility function of traffic participants, α ij represents the adjustment factor, which is the impact of traffic flow on the decision of each participant, V ij represents the traffic speed, β ij The moderating factor representing the disorder, D oij represents the impact of obstacles on the path, γ ij represents the interaction factor, μ ij represents the membership value of traffic participants, λ ij An adjustment factor representing the path selection priority; S43. According to the utility function of traffic participants, solve the game equilibrium. Assume that each traffic participant chooses strategy d k The impact on utility has a nonlinear relationship, and a nonlinear equation is used to solve the optimal strategy for each participant: Among them, U i (d k ) indicates the U i Traffic participants make decisions in the choice k The utility when c represents the number of decisions, f k represents the weight of the jth feature, m represents the number of features, and w j represents the weight of the jth feature; S44. Calculate the game optimization value of each candidate path based on the behavior and decision of each participant. Let G k is the candidate path P k The game optimization value of the candidate path P k Feasibility and priority after considering the behavior of traffic participants: Among them, G k Represents the candidate path P k The game optimization value, α i represents the influence factor of the ith traffic participant on the candidate route selection, U i (p i ) indicates that the i-th traffic participant selects the candidate path P k The utility value at time θ i represents the adjustment coefficient that affects the decision-making of participants, γ ij represents the interaction factor, μ ij It indicates the membership value of traffic participants when choosing a path.

5. The intelligent driving multi-objective path planning method based on fuzzy game optimization according to claim 1 is characterized in that: The S5 specifically includes: S51, combine the optimization target weight after fuzzy logic processing with the game optimization value, and set W i is the weight of the i-th optimization objective, G k is the candidate path P k The game optimization value of each candidate path P k , according to the optimization target weight and game optimization value after fuzzy logic processing, the comprehensive evaluation value of each path is calculated: Among them, C k represents the comprehensive evaluation value, W i represents the weight of the i-th optimization objective, O ki Represents the path P k The normalized value of the corresponding i-th optimization objective, β represents the adjustment coefficient of the game optimization value, G k Represents the candidate path P k The game optimization value of ζ represents the interaction coefficient, interaction(P k ) represents the candidate path P k The interaction effect with traffic participants is the mutual influence between route choices; S52, using the improved particle swarm optimization algorithm to perform multi-objective optimization on the candidate paths, setting the search space of the improved particle swarm optimization algorithm to S = {P1, P2, ..., P m }, P in the search space m represents the mth candidate path, m represents the number of paths, the search space of each candidate path is a particle, and each particle P k The state of is represented by position and velocity, and each particle is iteratively updated by the improved particle swarm optimization algorithm, which includes inertia, cognitive and social terms: in, Represents particle P k The speed at the t+1th iteration, Represents particle P k The speed at the tth iteration, Represents particle P k The position at the tth iteration, Represents particle P k The position in the t+1th iteration, P best Represents particle P k The personal optimal position, G best represents the global optimal position, c1 and c2 represent learning factors, rand1 and rand2 represent random numbers, and ω represents the adaptive inertia weight, which controls the particle speed update; S53, introduce adaptive inertia weight ω to dynamically adjust the exploration ability of particles. The inertia weight decreases with the number of iterations: Where ω(t) represents the adaptive inertia weight in the tth iteration, ω max represents the initial inertia weight, ω min represents the minimum inertia weight, T represents the maximum number of iterations, and t represents the current number of iterations; S54, gradually adjust the path selection strategy during the search process, and the improved particle swarm optimization algorithm is optimized through global and local search to update the comprehensive evaluation value C k : in, Represents particle P k The position at the t+1th iteration, Represents particle P k The position in the tth iteration, λ represents the step size factor, P best Represents particle P k The personal optimal position of the particle, μ represents the local search step factor, which adjusts the search amplitude of the particle near the local optimal solution. Represents the path P k The comprehensive evaluation value after the tth iteration, Represents particle P k The gradient of the comprehensive evaluation value; S55. Through the combined effect of multiple search processes and local search mechanisms, the overall performance of candidate paths is gradually improved to generate the optimal path: Among them, P opt represents the optimal path, Represents the path P k The final comprehensive evaluation value after the Tth iteration, S represents the search space of the improved particle swarm optimization algorithm, and argmax represents the variable value when the function takes the maximum value.

Images