Vehicle speed planning method and system based on cloud dynamic optimization

Abstract

The invention provides a vehicle speed planning method and system based on cloud dynamic optimization, and belongs to the technical field of vehicle speed planning, and the method comprises the steps that a cloud server obtains the current state of a vehicle and the front road information of the vehicle, and carries out the road section division of the front road of the vehicle; the cloud server constructs a multi-objective optimization model by taking the shortest driving time and the least parking frequency of the vehicle on the road ahead as optimization objectives and taking the vehicle speed of the vehicle on each road section as a decision variable, and solves the planned vehicle speed of each road section as an optimal solution by using a multi-objective particle swarm optimization algorithm; and the cloud server returns the planned speed of the real-time road section to which the vehicle belongs to the vehicle to guide the driver to drive. The vehicle speed is optimized through real-time interaction between the vehicle and the cloud, and the driving efficiency is improved. The vehicle speed is planned by adopting the multi-target particle swarm algorithm, the parking frequency and the driving time are reduced, and accurate vehicle speed planning is provided for the vehicle.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle speed planning, and particularly relates to a vehicle speed planning method and system based on cloud dynamic optimization. Background Art

[0002] With the development of intelligent transportation systems and the progress of autonomous driving technologies, vehicle speed planning plays a crucial role in enhancing driving safety, promoting energy conservation and emission reduction, and improving road traffic efficiency. However, traditional vehicle speed planning methods are usually based on static models, ignoring dynamic traffic environments and road conditions, such as the state transition of traffic signals, the undulation of road gradients, the adjustment of speed limits, and the fluctuations of real-time road conditions and other core elements. Therefore, traditional vehicle speed planning lacks real-time response to these dynamic core elements, ultimately resulting in the difficulty of the planned vehicle speed to match the actual situation, thereby leading to problems such as low traffic efficiency, increased energy consumption, and potential driving safety hazards.

[0003] In related patents, there have been some vehicle speed planning technologies that have optimized vehicle speed planning to a certain extent, but there are still limitations. Specifically, the existing vehicle speed planning fails to comprehensively and deeply integrate real-time traffic information and road conditions, especially this key signal of traffic signals, for vehicle speed planning. Therefore, in the face of complex and changing traffic environments, both traditional vehicle speed planning and the new generation of vehicle speed planning are difficult to meet the requirements of accurate vehicle speed planning in complex traffic scenarios, and the limitations are more prominent in urban roads with traffic congestion and frequent signal changes, greatly affecting vehicle speed driving efficiency and energy utilization efficiency. Summary of the Invention

[0004] In a first aspect, an embodiment of the present application provides a vehicle speed planning method based on cloud dynamic optimization, including the following steps: S1. The cloud server obtains the current state of the vehicle and the road information in front of the vehicle, and divides the road in front of the vehicle into sections; S2. The cloud server constructs a multi-objective optimization model with the shortest driving time and the fewest parking times of the vehicle on the road in front as the optimization objectives, and the vehicle speed in each section as the decision variable, and uses the multi-objective particle swarm optimization algorithm to solve the planned vehicle speed of each section as the optimal solution; S3. The cloud server returns the planned vehicle speed of the real-time section to which the vehicle belongs to the vehicle to guide the driver's driving.

[0005] Further, the specific steps of step S1 are as follows: S11. The vehicle obtains the current state of the vehicle through an in-vehicle GPS system, and the current state of the vehicle includes the vehicle position and the vehicle speed; S12. The vehicle obtains the road ahead through an in-vehicle navigation system; S13. The vehicle sends the current vehicle state and the road ahead to the cloud server through the communication module; S14. The cloud server obtains the road information ahead according to the road ahead in combination with the cloud map service and the road traffic system; the road information ahead includes the speed limit signs of road sections, the positions of traffic lights, and the states of traffic lights; S15. The cloud server divides the total journey of the vehicle's road ahead into several road sections according to the speed limit signs of road sections and the positions of traffic lights.

[0006] Further, the specific steps of step S2 are as follows: S21. The cloud server sets the shortest driving time of the vehicle on the road ahead as the first goal and the least number of stops of the vehicle on the road ahead as the second goal; S22. The cloud server constructs a multi-objective optimization model with the vehicle speed on the road section as the decision variable and the speed limits of each road section as the constraints:

[0007] Among them, represents the optimization objective vector, x = x 1, x 2,…, x D is the planned vehicle speed vector, x 1 is the vehicle speed of the first road section, x D is the vehicle speed of the D th road section, Ω is the constraint composed of the speed limits of each road section, T total (x) represents the driving time for the vehicle to complete the total journey, which is the first optimization goal; S total (x) represents the number of stops for the vehicle to complete the total journey, which is the second optimization goal; S23. Use the multi-objective particle swarm optimization algorithm to solve a set of optimal planned vehicle speed vectors as the optimal solution.

[0008] Further, the specific steps of step S23 are as follows: S231. Randomly generate particle positions and velocities with the vehicle speed planning scheme as the particle, take the randomly generated particle positions as the initial particle positions, and calculate the optimization objective vector ; S232. Take the initial particle positions as the initial individual optimal solutions, establish an external archive to store non-dominated solutions based on the dominance relationship of each optimization objective in the optimization objective vector of the initial particle positions, and randomly select the global optimal solution from the external archive; S233. Construct the convergence index and diversity index of non-dominated solutions, evaluate the exploitation performance and exploration performance of the population according to the convergence index and diversity index, then judge the population evolution state, and design the scaling factor according to the population evolution state; S234. Use the individual optimal trajectory prediction model based on knowledge reuse, combine the scaling factor to guide the population evolutionary search, predict the individual optimal solution in the next iteration process, and randomly select the global optimal solution from the external archive; S235. Update the velocity and position of the particles according to the predicted individual optimal solution and the selected global optimal solution; S236. Integrate the individual optimal solution with the external archive, select the non-dominated solutions among them to update the external archive; S237. Judge whether the iteration process of population evolution satisfies the maximum number of iterations; If yes, go to step S239; If not, go to step S238; S238. Enter the next iteration process of population evolution, and return to step S233; S239. Output the global optimal solution as the optimal planned vehicle speed vector.

[0009] Furthermore, the specific steps of step S232 are as follows: S2321. Use the planned vehicle speeds of each section in each vehicle speed planning scheme corresponding to the initial particle positions as the initial individual optimal solutions; S2322. Calculate the optimization objective vector based on the initial individual optimal solutions, and obtain two optimization objectives corresponding to each vehicle speed planning scheme; S2323. Establish an external archive; S2324. Compare the dominance relationships of the driving times and parking times of each vehicle speed planning scheme, and the vehicle speed planning schemes that are not dominated by any other solutions are used as non-dominated solutions and saved to the external archive; S2325. Randomly select a vehicle speed planning scheme from the external archive as the global optimal solution.

[0010] Furthermore, the specific steps of step S233 are as follows: S2331. Construct the convergence index of the optimal solution in the population iteration process through the following formula:

[0011] where \(C(t)\) is the convergence index of the \(t\)-th generation, \(m = 1, 2\) is the number of objectives, \(A(t)\) is the non-dominated solution set in the external archive, \(|A(t)|\) is the number of non-dominated solutions in the external archive, is the \(m\)-th objective value of the \(n\)-th non-dominated solution in the external archive in the \((t - 1)\)-th iteration process; S2332. Construct the diversity index of the optimal solution during the population iteration process through the following formula:

[0012]

[0013] where D ( t ) is the diversity index of the t th generation, S ( t ) is the distribution performance of the external archive in the t th iteration, d n ( t ) is the crowding distance of the n th non-dominated solution in the objective space in the external archive, is the average value of the crowding distances of all non-dominated solutions in the external archive; S2333. Set the development performance to characterize the performance of the population evolving by searching for a new vehicle speed planning scheme based on the existing vehicle speed planning scheme, and set the exploration performance to characterize the performance of the population evolving by optimizing the vehicle speed of some road sections of the existing vehicle speed planning scheme; S2334. Judge the evolutionary state of the population; When C ( t ) > 0 and D ( t ) > 0, it is determined that both the development performance and the exploration performance of the population meet the requirements, and the current evolutionary mode of the population is maintained; When C ( t ) ≥ 0 and D ( t ) < 0, it is determined that the exploration performance of the population is low, and the self-evolution of the particles is further guided; When C ( t ) = 0 and D ( t ) > 0, it is determined that the development performance of the population is low, and the scope of population evolution is further expanded; S2335. Design a scaling factor , and set that the exploration performance of the population is enhanced when increases, and set that the development performance of the population is enhanced when decreases.

[0014] Furthermore, the specific steps of step S234 are as follows: S2341. Set the historical information of the individual optimal solution constructed with the knowledge K ( T ) obtained during the population iteration processP i,d ( T );

[0015]

[0016] Among them, K ( T ) is the knowledge obtained by the population in the t th iteration process, P i,d ( T ) is the historical information composed of the optimal solutions of the i th individual at different times, T = t-n + 1, t-n + 2, …, t is the matrix composed of the historical iteration times, n is a preset parameter; S2342. Construct the cost function of the individual optimal trajectory prediction model based on knowledge reuse;

[0017] Among them, is the parameter matrix, k is the number of samples, is the regression equation, expressed as follows:

[0018] Among them, is the designed scaling factor, designed based on the adaptive dynamic feedback mechanism; S2343. Let the cost function be the minimum, and use the least squares method to solve the parameter matrix : ; Further derivation gives: ; Further simplification gives:

[0019] Let , and the solution can be obtained: ; S2344. Predict the individual optimal solution in the next iteration process according to the obtained parameter matrix: ; S2345. Judge whether dominates ; If yes, go to step S2346; If no, go to step S2347; S2346. Take as t the individual optimal solution of the i th particle of the +1 generation, and go to step S2348; = as t the individual optimal solution of the i th particle of the +1 generation; In step S235, update the velocity and position of the particle according to the individual optimal solution and the global optimal solution using the following formula:

[0020]

[0021] where, v i,d ( t ) and x i,d ( t ) represent the velocity and position of the t th iteration of the i th particle in the d th dimension, i = 1, 2, … N is the number of particles; d = 1, 2, …, D , p i,d ( t ) is the individual optimal solution of the g d ( t ) is the global optimal solution searched by the population, w is the inertia weight, c 1 and c 2 are the individual cognitive factor and the social cognitive factor respectively, r 1 and r 2 are random numbers in the range [0, 1].

[0022] Furthermore, the specific steps of step S236 are as follows: S2361. Determine whether the individual optimal solution is dominated by the vehicle speed planning scheme in the external file; If yes, discard the individual optimal solution and go to step S237; Otherwise, store the individual optimal solution in the external archive, and delete the vehicle speed planning schemes in the external archive that are dominated by the individual optimal solution; S2362. Determine whether the number P of vehicle speed planning schemes that are non-dominated solutions in the external archive is greater than the maximum size N of the external archive; If so, proceed to step S2363; If not, proceed to step S237; S2363. Sort each non-dominated solution in the external archive according to the crowding distance, and delete the non-dominated solution with the smallest crowding distance, then return to step S2362.

[0023] Furthermore, the specific steps of step S3 are as follows: S31. The cloud server associates each planned vehicle speed in the optimal planned vehicle speed vector with the corresponding road segment; S32. The cloud server identifies the real-time road segment to which the vehicle belongs, and returns the planned vehicle speed corresponding to the real-time road segment to the vehicle to guide the driver's driving; S33. The cloud server identifies whether the vehicle enters a new road segment; If so, re-perform vehicle speed planning and return to step S1; If not, proceed to step S34; S34. Wait for a set time period and return to step S33.

[0024] In a second aspect, an embodiment of the present application further provides a vehicle speed planning system based on cloud dynamic optimization, including: A data acquisition and road segment division module, configured to obtain the current state of the vehicle and the road information in front of the vehicle on the cloud server, and divide the road in front of the vehicle into road segments; A vehicle speed planning module, configured to build a multi-objective optimization model with the shortest driving time and the fewest parking times of the vehicle on the front road as the optimization objectives and the vehicle speed in each road segment as the decision variables on the cloud server, and use the multi-objective particle swarm optimization algorithm to solve the planned vehicle speed of each road segment as the optimal solution; A planned vehicle speed feedback module, configured to return the planned vehicle speed of the real-time road segment to which the vehicle belongs to the vehicle on the cloud server to guide the driver's driving.

[0025] From the above technical solutions, it can be seen that the present invention has the following advantages: In the speed planning method and system based on cloud dynamic optimization provided by this application, through the real-time data interaction between the vehicle terminal and the cloud, the optimal speed can be calculated and adjusted in real time according to the changing traffic conditions and road environments, realizing the dynamic optimization of the speed and improving the vehicle driving efficiency; the multi-objective particle swarm optimization algorithm is adopted to plan the speed with the goal of minimizing the driving time and the number of stops, effectively reducing unnecessary stops and driving time, improving the energy utilization efficiency, and optimizing the driving process; the scaling factor adjustment mechanism based on dynamic feedback and the individual optimal trajectory prediction model based on knowledge reuse are introduced to adjust the exploration and development capabilities in the particle swarm optimization process in real time according to the convergence and diversity indexes of the population, improving the efficiency and accuracy of speed planning, avoiding falling into local optimal solutions, and providing accurate speed planning for vehicle driving; through the real-time collaborative work between the vehicle terminal and the cloud platform, the vehicle state data is uploaded to the cloud in real time, and the cloud calculates the optimized speed in real time according to the latest information and transmits it back to the vehicle terminal, forming a closed-loop feedback mechanism to ensure the timeliness and accuracy of speed planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0027] Figure 1 It is a schematic flow chart of the speed planning method based on cloud dynamic optimization of the present invention.

[0028] Figure 2 It is a schematic diagram of the speed planning system based on cloud dynamic optimization of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] In the following, the specific steps of the speed planning method based on cloud dynamic optimization will be described in detail, and various embodiments of the present disclosure will be described more comprehensively. The present disclosure can have various embodiments, and adjustments and changes can be made therein. However, it should be understood that there is no intention to limit the various embodiments of the present disclosure to the specific embodiments disclosed herein, but the present disclosure should be understood to cover all adjustments, equivalents, and / or alternative solutions falling within the spirit and scope of the various embodiments of the present disclosure.

[0030] Exemplarily, with the development of intelligent transportation systems and autonomous driving technologies, speed planning plays an increasingly important role in enhancing driving safety, promoting energy conservation and emission reduction, and improving road traffic efficiency. However, most traditional speed planning strategies are based on static models and ignore the dynamic characteristics of the traffic environment and road conditions, such as key factors like the state changes of traffic lights, the dynamic changes in road gradients, the immediate adjustment of speed limits, and the fluctuations in real-time traffic conditions. This results in the inability of traditional methods to respond to these dynamic elements in real time, causing the planned speeds to often deviate from the actual road conditions, thereby leading to problems such as low traffic efficiency, increased energy consumption, and driving safety risks.

[0031] Although some patents on speed planning technologies have improved speed planning to a certain extent, they still face limitations. Specifically, these technologies fail to fully and deeply integrate real-time traffic information and road conditions, especially the crucial traffic lights, for precise speed planning. Therefore, in the face of complex and changing traffic scenarios, neither traditional speed planning nor the new generation of speed planning can meet the demand for accurate speed planning. Especially in urban traffic, in the face of traffic congestion and frequent changes in traffic lights, the limitations of these methods are more significant, severely restricting the improvement of speed driving efficiency and energy utilization efficiency.

[0032] To address the above problems, this embodiment provides a speed planning method based on cloud dynamic optimization, which can handle complex traffic conditions and road condition changes, achieve real-time optimization and adjustment of vehicle speeds, thereby improving vehicle driving efficiency, reducing energy consumption, and optimizing the driving process.

[0033] Next, the technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0034] Please refer to Figure 1 The following shows a flowchart of the speed planning method based on cloud dynamic optimization in a specific embodiment. The method includes the following steps: S1. The cloud server obtains the current state of the vehicle and the road information in front of the vehicle, and divides the road in front of the vehicle into sections; It should be noted that the cloud server obtains the current state of the vehicle and the road information in front, enabling the cloud to comprehensively understand the basic information of vehicle driving, providing a data basis for speed planning. Dividing the road in front into sections decomposes the complex road journey into multiple sub-sections, facilitating refined speed planning for each sub-section and improving the accuracy of speed planning; S2. The cloud server constructs a multi-objective optimization model with the shortest driving time and the fewest stops of the vehicle on the forward road as the optimization objectives, and uses the vehicle speed on each section as the decision variable, and solves the planned vehicle speed of each section as the optimal solution using the multi-objective particle swarm optimization algorithm; It should be noted that by taking the shortest driving time and the fewest stops as the optimization objectives, constructing a multi-objective optimization model, and introducing two key factors in vehicle driving, the driving efficiency of the vehicle can be effectively improved, and energy consumption and driving time can be reduced; while using the multi-objective particle swarm optimization algorithm to solve the planned vehicle speed can find the optimal vehicle speed planning scheme in the complex solution space and improve the accuracy and efficiency of vehicle speed planning; S3. The cloud server returns the planned vehicle speed of the real-time section to which the vehicle belongs to the vehicle to guide the driver's driving; It should be noted that returning the planned vehicle speed of the real-time section to the vehicle to guide the driver's driving enables the driver to drive at the optimized vehicle speed, realizes the guiding effect of vehicle speed planning on actual driving, improves the safety and efficiency of driving, and re-plans the vehicle speed according to whether the vehicle enters a new section, ensuring that the vehicle speed planning can adapt to road changes in real time and enhancing the rationality of vehicle speed suggestions.

[0035] This embodiment realizes the dynamic and intelligent planning of vehicle speed by integrating the vehicle's real-time state, forward road information and the cloud server, improving driving safety, energy conservation and emission reduction effects, and road traffic efficiency.

[0036] Further, as a refinement and extension of the specific implementation manner of the above embodiment, in order to fully illustrate the specific implementation process in this embodiment, another vehicle speed planning method based on cloud dynamic optimization is provided, and this method includes the following steps: S1. The cloud server obtains the current state of the vehicle and the forward road information of the vehicle, and divides the forward road of the vehicle into sections; The specific steps of step S1 are as follows: S11. The vehicle obtains the current state of the vehicle through the in-vehicle GPS system, and the current state of the vehicle includes the vehicle position and the vehicle speed; S12. The vehicle obtains the forward driving road through the in-vehicle navigation system; S13. The vehicle sends the current state of the vehicle and the forward driving road to the cloud server through the communication module; S14. The cloud server obtains the forward road information according to the forward driving road in combination with the cloud map service and the road traffic system; The forward road information includes section speed limit signs, traffic signal positions and traffic signal states; S15. The cloud server divides the total journey of the vehicle's forward road into several sections according to the section speed limit signs and traffic signal positions; S2. The cloud server constructs a multi-objective optimization model with the shortest driving time and the fewest parking times of the vehicle on the forward road as the optimization objectives, and uses the multi-objective particle swarm optimization algorithm to solve the planned vehicle speed of each section as the optimal solution. The specific steps of step S2 are as follows: S21. The cloud server sets the shortest driving time of the vehicle on the forward road as the first objective and the fewest parking times of the vehicle on the forward road as the second objective; S22. The cloud server constructs a multi-objective optimization model with the vehicle speed on the section as the decision variable and the speed limit of each section as the constraint:

[0037] where, represents the optimization objective vector, x = x 1, x 2,…, x D is the planned vehicle speed vector, x 1 is the vehicle speed of the first section, x D is the vehicle speed of the D th section, Ω is the constraint composed of the speed limits of each section, T total (x) represents the driving time for the vehicle to complete the total journey, which is the first optimization objective; S total (x) represents the number of parking times for the vehicle to complete the total journey, which is the second optimization objective; It should be noted that the status of traffic lights can also be added as a constraint condition; S23. Use the multi-objective particle swarm optimization algorithm to solve a set of optimal planned vehicle speed vectors as the optimal solution. The specific steps of step S23 are as follows: S231. Randomly generate particle positions and velocities with the vehicle speed planning scheme as the particle, take the randomly generated particle position as the initial particle position, and calculate the optimization objective vector ; S232. Take the initial particle position as the initial individual optimal solution, establish an external archive to store non-dominated solutions based on the dominance relationship of each optimization objective in the optimization objective vector of the initial particle position, and randomly select the global optimal solution from the external archive; It should be noted that in speed planning, each particle represents a possible speed planning scheme. The speed planning corresponding to the self-optimal state experienced by the particle during the search process is the individual optimal solution. For example, assume there is a journey consisting of 5 sections. Particle A obtained a speed planning scheme during the previous search, which performed well in terms of shorter target driving time and fewer stops. The speeds corresponding to each section in this scheme constitute the individual optimal solution of Particle A. The individual optimal solution is the optimal speed planning record in the particle's own history and is not necessarily the optimal solution in the entire population. The global optimal solution refers to the optimal speed planning scheme searched by all particles in the entire population. Taking the above journey as an example again, when all particles in the population have undergone multiple iterative searches and found a speed planning scheme that minimizes the driving time and the number of stops under the constraints such as section speed limits, the speeds corresponding to each section are the global optimal solution. The global optimal solution is the most ideal speed planning found by the current population during the search process and is used to guide vehicle driving. S233. Construct the convergence index and diversity index of non-dominated solutions, evaluate the exploitation performance and exploration performance of the population based on the convergence index and diversity index, then judge the population evolution state, and design the scaling factor according to the population evolution state. S234. Use the individual optimal trajectory prediction model based on knowledge reuse, combine the scaling factor to guide the population evolution search, predict the individual optimal solution in the next iteration process, and randomly select the global optimal solution from the external archive. It should be noted that the individual optimal trajectory prediction model based on knowledge reuse utilizes the knowledge obtained by the population in historical iterations (i.e., the historical information composed of individual optimal solutions at different times) to predict the evolution trend of the population. In speed planning, by analyzing the past speed planning schemes as knowledge, it is possible to predict the possible better future speed planning directions. For example, if it is found that the speed adjustment pattern in certain sections in the past always brings better target results, then in subsequent iterations, this knowledge can be used to guide the search direction of the particle for dynamic feedback adjustment, enabling the particle to search in the direction more likely to find the optimal speed planning. S235. Update the speed and position of the particle according to the predicted individual optimal solution and the selected global optimal solution. It should be noted that the speed of the particle is the moving speed of the particle in the search space in the multi-objective particle swarm optimization algorithm. In speed planning, the speed of the particle determines the adjustment amplitude of the speed planning scheme by the particle in the next iteration. For example, when the speed of the particle is relatively large, the particle may make a large change to the current speed planning scheme and try new speed combinations. When the speed is small, the adjustment amplitude is small, and it is more inclined to fine-tune near the current scheme. The position of a particle is the state of the particle in the search space, corresponding to a vehicle speed planning scheme; in the vehicle speed planning scenario, the position of each particle consists of the vehicle speeds planned for each sub-section; for example, if the vehicle speeds planned by a particle on 5 sub-sections are [30, 40, 35, 45, 38] (unit: km / h), this set of vehicle speeds represents the position of the particle, that is, a specific vehicle speed planning scheme; each time the position is updated, the vehicle speeds of each section are adjusted according to the updated speed to form a new vehicle speed planning scheme; S236. Integrate the individual optimal solution with the external archive, and select the non-dominated solutions among them to update the external archive; S237. Determine whether the iterative process of population evolution meets the maximum number of iterations; If so, go to step S239; If not, go to step S238; S238. Enter the next iterative process of population evolution and return to step S233; S239. Output the global optimal solution as the best planned vehicle speed vector; S3. The cloud server returns the planned vehicle speeds of the real-time section to which the vehicle belongs to the vehicle to guide the driver's driving; the specific steps of step S3 are as follows: S31. The cloud server corresponds each planned vehicle speed in the best planned vehicle speed vector to the corresponding section; S32. The cloud server identifies the real-time section to which the vehicle belongs and returns the planned vehicle speed corresponding to the real-time section to the vehicle to guide the driver's driving; S33. The cloud server identifies whether the vehicle enters a new section; If so, re-plan the vehicle speed and return to step S1; If not, go to step S34; S34. Wait for the set time period and return to step S33.

[0038] In an embodiment of the present invention, based on step S232, step S233, step S234, step S2365, and step S236, a possible embodiment will be given below to non-limitingly elaborate on its specific implementation scheme.

[0039] The specific steps of step S232 are as follows: S2321. Use the planned vehicle speeds of each section in each vehicle speed planning scheme corresponding to the initial particle position as the initial individual optimal solution; S2322. Calculate the optimization objective vector based on the initial individual optimal solution , and obtain two optimization objectives corresponding to each vehicle speed planning scheme; S2323. Establish an external archive; S2324. Compare the dominance relationships of the driving times and the number of stops among different vehicle speed planning schemes. The vehicle speed planning scheme that is not dominated by any other solution is regarded as a non-dominated solution and saved to an external file. S2325. Randomly select a vehicle speed planning scheme from the external file as the global optimal solution. It should be noted that the dominance relationship is a relationship for comparing the advantages and disadvantages of different vehicle speed planning schemes. If one scheme is superior to another in all objectives, or has advantages in some objectives and is not inferior in other objectives, then it is said to dominate the other scheme. The non-dominated solutions are screened out through the dominance relationship, and then a better vehicle speed planning scheme can be found. For two vehicle speed planning schemes X and Y, if X is not worse than Y in terms of the driving time and the number of stops, which are the optimization objectives, and X is superior to Y in at least one objective, then X is said to dominate Y. If there is no such dominance relationship between X and Y, then they are both non-dominated solutions. The external file is a storage structure for saving the non-dominated solutions found by the population during the search process. In vehicle speed planning, the external file serves as a library of excellent vehicle speed planning schemes, storing different vehicle speed planning schemes that perform well in terms of the driving time and the number of stops, which are multi-objectives, and there is no dominance relationship among them. For example, if the driving time of scheme X is shorter than that of scheme Y, but the number of stops is more than that of Y, and scheme Y has no absolute advantage in other aspects, then both X and Y are non-dominated solutions and may be saved to the external file. The specific steps of step S233 are as follows: S2331. Construct the convergence index of the optimal solution during the population iteration process through the following formula:

[0040] where C(t) is the convergence index of the t-th generation, m = 1, 2 is the number of objectives, A(t) is the non-dominated solution set in the external file, |A(t)| is the number of non-dominated solutions in the external file, is the m-th objective value of the n-th non-dominated solution in the external file during the (t - 1)-th iteration process. S2332. Construct the diversity index of the optimal solution during the population iteration process through the following formula:

[0041]

[0042] where D ( t ) is the diversity index of the t -th generation, S ( t ) is the external file at the tThe distribution performance of the current iteration d n ( t ) is the crowding distance of the n -th non-dominated solution in the objective space in the external archive, and is the average value of the crowding distances of all non-dominated solutions in the external archive; It should be noted that in the population evolution of vehicle speed planning, the exploration performance refers to the ability of particles to discover new and potentially better vehicle speed planning solutions in the search space; for example, on a section of road, the vehicle speed range originally searched by the particles is concentrated in a certain interval, but through strong exploration performance, the particles can try different vehicle speed combinations, explore other possible vehicle speed intervals, and see if they can find a better balance between driving time and the number of stops; for example, try higher vehicle speeds on some sections to reduce driving time, and at the same time observe whether it will cause a significant increase in the number of stops, so as to discover a new and better vehicle speed planning solution; The development performance refers to the ability of particles to perform local optimization on the currently discovered better vehicle speed planning solutions; when the particles find a relatively good vehicle speed planning solution, for example, when the optimal vehicle speeds of some sections have been determined, the development performance can allow the particles to fine-tune the vehicle speeds of some sections on the basis of this solution, and further optimize objectives such as driving time and the number of stops; for example, in a known good vehicle speed plan, slightly adjust the vehicle speed of a certain section to see if the overall performance can be improved; S2334. Determine the evolution state of the population; When C ( t ) > 0 and D ( t ) > 0, it is determined that both the development performance and exploration performance of the population meet the requirements, and the current population evolution method is maintained; When C ( t ) ≥ 0 and D ( t ) < 0, it is determined that the exploration performance of the population is low, and the particle self-evolution is further guided; When C ( t ) = 0 and D ( t ) > 0, it is determined that the development performance of the population is low, and the scope of population evolution is further expanded; It should be noted that the evolutionary state of the population can be judged by the convergence and diversity indicators. When C(t)>0 and D(t)>0, it indicates that both the exploration ability and the exploitation ability of the population are improved, and the current evolutionary direction is beneficial to the population evolution. At this time, the original population evolution rule should be maintained; When C(t)≥0 and D(t)<0, it indicates that the convergence of the optimal solution is improved while the diversity becomes worse. To prevent the particle swarm from falling into the local optimum, the exploration ability of the particles should be improved at this time, and a better individual optimal solution should be provided for the particles to guide the evolution of their own particles; When C(t)=0 and D(t)>0, it indicates that the convergence of the external archive is not improved and the particle exploitation fails to obtain an effective solution. Therefore, to prevent the particles from moving away from the optimal solution, the exploitation ability of the particles should be improved at this time; S2335. Design the scaling factor , and set the exploration performance of the population is enhanced when it increases, and set the exploitation performance of the population is enhanced when it decreases; Exemplarily, it can be set ; When increases, predict the individual optimal solution of the next steps with the individual optimal solution of a larger step size as the optimal solution of the current iteration to improve the exploration performance of the population; when decreases, the prediction of the individual optimal solution is more inclined to select around the current optimal solution, thereby improving the exploitation performance of the population; In vehicle speed planning, it is a parameter dynamically adjusted according to the evolutionary state of the population (judged by the convergence and diversity indicators); when both the exploration performance and the exploitation performance of the population are improved (C(t)>0 and D(t)>0), the scaling factor remains unchanged and the original search strategy is maintained; when the convergence is improved but the diversity becomes worse (C(t)≥0 and D(t)<0), the scaling factor is increased, and the particles are allowed to predict the individual optimal solution of the next step with the individual optimal solution of a larger step size, so as to improve the exploration performance and find more different vehicle speed planning schemes to avoid falling into the local optimum; for example, originally the particles search for a better vehicle speed planning in a certain local area, but may miss other better schemes. After increasing the scaling factor, the particles can expand the search range; when the convergence is not improved but the diversity is good (C(t)=0 and D(t)>0), the scaling factor is decreased, so that the prediction of the individual optimal solution is more inclined to select around the current optimal solution, improving the exploitation performance and finely adjusting the better vehicle speed planning found currently; The specific steps of step S234 are as follows: S2341. Set the knowledge obtained during the population iteration process K (T )Construct the historical information of the individual optimal solution P i,d ( T );

[0043]

[0044] Among them, K ( T ) is the knowledge obtained by the population in the t -th iteration process, P i,d ( T ) is the historical information composed of the individual optimal solutions of the i -th individual at different times, T = t-n + 1, t-n + 2, …, t ) is the matrix composed of the historical iteration times, n is the preset parameter; S2342. Construct the cost function of the individual optimal trajectory prediction model based on knowledge reuse;

[0045] Among them, is the parameter matrix, k is the number of samples, is the regression equation, which is expressed as follows:

[0046] Among them, is the designed scaling factor, which is designed based on the adaptive dynamic feedback mechanism; S2343. Let the cost function be minimized, and use the least squares method to solve the parameter matrix : ; Further derivation gives: ; Then further simplify to get:

[0047] Let , and the solution can be obtained: ; S2344. Predict the individual optimal solution in the next iteration process according to the obtained parameter matrix: S2345. Judge whether it satisfies Whether ; If yes, go to step S2346; If no, go to step S2347; S2346. Take as t the individual optimal solution of the i -th generation of the particle, and go to step S2348; = as t the individual optimal solution of the i -th generation of the particle; In step S235, update the velocity and position of the particle according to the individual optimal solution and the global optimal solution using the following formula:

[0048]

[0049] where, v i,d ( t ) and x i,d ( t ) represent the velocity and position of the t -th iteration of the i -th particle in the d -th dimension, i = 1, 2, … N is the number of particles, N is set to 100; d = 1, 2, …, D , p i,d ( t ) is the individual optimal solution of the g d ( t ) is the global optimal solution found by the population, w is the inertia weight, c 1 and c 2 are the individual cognitive factor and the social cognitive factor respectively, r 1 and r 2 are random numbers in the range [0, 1]; The specific steps of step S236 are as follows: S2361. Judge whether the individual optimal solution is dominated by the vehicle speed planning scheme in the external archive; If yes, discard the individual optimal solution and go to step S237; Otherwise, store the individual optimal solution in the external archive and delete the vehicle speed planning schemes in the external archive that are dominated by the individual optimal solution; S2362. Determine whether the number P of vehicle speed planning schemes that are non-dominated solutions in the external archive is greater than the maximum size N of the external archive; If so, proceed to step S2363; If not, proceed to step S237; S2363. Sort each non-dominated solution in the external archive according to the crowding distance, and delete the non-dominated solution with the smallest crowding distance, then return to step S2362; Exemplarily, taking the case where there are 10 particles in the population as an example, after one iteration, each particle generates a vehicle speed planning scheme; after judging the domination relationship, it is found that 6 of them are non-dominated solutions; at this time, there are already 80 non-dominated solutions in the external archive (assuming 80 have been stored before), and after adding these 6, the number of non-dominated solutions becomes 86, which is less than the maximum size of 100, so these 6 non-dominated solutions are directly stored in the external archive; After the next iteration, new vehicle speed planning schemes are generated, and after judgment, there are 15 non-dominated solutions; at this time, the number of non-dominated solutions in the external archive becomes 86 + 15 = 101, exceeding the maximum size of 100; at this time, it is necessary to sort all the non-dominated solutions in the external archive according to the crowding distance, delete the solution with the smallest crowding distance, and then judge whether the number still exceeds 100. If it still exceeds, continue to delete the solution with the smallest crowding distance until the number of non-dominated solutions is 100; update the external archive in this way to ensure that the stored are non-dominated solutions and the number meets the set requirements, providing a basis for the selection of the subsequent global optimal solution.

[0050] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of the present invention.

[0051] As Figure 2 shown, the following is an embodiment of a vehicle speed planning system based on cloud dynamic optimization provided by an embodiment of the present disclosure. This system and the vehicle speed planning method based on cloud dynamic optimization in the above embodiments belong to the same inventive concept. For the details not described in detail in the embodiment of the vehicle speed planning system based on cloud dynamic optimization, reference can be made to the embodiment of the vehicle speed planning method based on cloud dynamic optimization.

[0052] The system includes: A data acquisition and road section division module, configured to obtain the current state of the vehicle and the road information in front of the vehicle on the cloud server, and divide the road in front of the vehicle into road sections; The vehicle speed planning module is used to construct a multi-objective optimization model with the shortest driving time and the fewest parking times of the vehicle on the forward road as the optimization objectives in the cloud server, and use the multi-objective particle swarm optimization algorithm to solve the planned vehicle speed of each section as the optimal solution, taking the vehicle speed of each section as the decision variable; The planned vehicle speed feedback module is used to return the planned vehicle speed of the real-time section to which the vehicle belongs to the vehicle in the cloud server to guide the driver's driving.

[0053] In this embodiment, the automation and intelligence of vehicle speed planning are realized through the data acquisition and section division module, the vehicle speed planning module, and the planned vehicle speed feedback module.

[0054] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A vehicle speed planning method based on cloud dynamic optimization, characterized in that, It includes the following steps: S1. The cloud server obtains the current state of the vehicle and the information of the road ahead of the vehicle, and divides the road ahead of the vehicle into sections; S2. The cloud server constructs a multi-objective optimization model with the shortest driving time of the vehicle on the road ahead and the least number of stops as the optimization objectives, and uses the multi-objective particle swarm optimization algorithm to solve the planned vehicle speed of each section as the optimal solution; S3. The cloud server returns the planned vehicle speed of the real-time section to which the vehicle belongs to the vehicle to guide the driver's driving.

2. The vehicle speed planning method based on cloud dynamic optimization according to claim 1, characterized in that The specific steps of step S1 are as follows: S11. The vehicle obtains the current state of the vehicle through the in-vehicle GPS system, and the current state of the vehicle includes the vehicle position and the vehicle speed; S12. The vehicle obtains the road ahead through the in-vehicle navigation system; S13. The vehicle sends the current state of the vehicle and the road ahead through the communication module to the cloud server; S14. The cloud server obtains the road information ahead according to the road ahead in combination with the cloud map service and the road traffic system; the road information ahead includes the section speed limit sign, the position of the traffic signal and the state of the traffic signal; S15. The cloud server divides the total journey of the road ahead of the vehicle into several sections according to the section speed limit sign and the position of the traffic signal.

3. The vehicle speed planning method based on cloud - side dynamic optimization according to claim 2, wherein, The specific steps of step S2 are as follows: S21. The cloud server sets the shortest driving time of the vehicle on the road ahead as the first objective and the least number of stops of the vehicle on the road ahead as the second objective; S22. The cloud server constructs a multi-objective optimization model with the vehicle speed in the section as the decision variable and the speed limit of each section as the constraint: Among them, represents the optimization target vector, \(x = x 1, x 2, \ldots, x D is the planned speed vector, x v_1\) is the speed of the first section, x D is the speed of the D -th section, \(\Omega\) is the constraint composed of the speed limits of each section, T total T(x)\) represents the driving time for the vehicle to complete the total journey, which is the first optimization target; S total P(x)\) represents the number of stops for the vehicle to complete the total journey, which is the second optimization target; S23. Use the multi-objective particle swarm optimization algorithm to solve a set of optimal planned vehicle speed vectors as the optimal solution.

4. The vehicle speed planning method based on cloud dynamic optimization according to claim 3, wherein The specific steps of step S23 are as follows: S231. Randomly generate particle positions and velocities based on the vehicle speed planning scheme, use the randomly generated particle positions as the initial particle positions, and calculate the optimization objective vector ; S232. Take the initial particle position as the initial individual optimal solution, and establish an external archive to store non-dominated solutions based on the domination relationship of each optimization objective in the optimization objective vector of the initial particle position. Randomly select the global optimal solution from the external archive; ​ S233. Construct the convergence index and diversity index of the non-dominated solutions, evaluate the exploitation performance and exploration performance of the population according to the convergence index and diversity index, then judge the population evolution state, and then design the scaling factor according to the population evolution state; S234. Use the individual optimal trajectory prediction model based on knowledge reuse, combine the scaling factor to guide the population evolution search, predict the individual optimal solution in the next iteration process, and randomly select the global optimal solution from the external archive; S235. Update the speed and position of the particle according to the predicted individual optimal solution and the selected global optimal solution; S236. Integrate the individual optimal solution with the external archive, and select the non-dominated solutions among them to update the external archive; S237. Judge whether the iteration process of population evolution meets the maximum number of iterations; If so, go to step S239; If not, go to step S238; S238. Enter the next iteration process of population evolution and return to step S233; S239. Output the global optimal solution as the optimal planned vehicle speed vector.

5. The vehicle speed planning method based on cloud dynamic optimization according to claim 4, characterized in that The specific steps of step S232 are as follows: S2321. Take the planned vehicle speed of each section in each vehicle speed planning scheme corresponding to the initial particle position as the initial individual optimal solution; S2322. Calculate the optimization objective vector based on the initial individual optimal solution , and obtain two optimization objectives corresponding to each vehicle speed planning scheme; S2323. Establish an external archive; S2324. Compare the dominance relationships of the driving times and the number of stops among the vehicle speed planning schemes. The vehicle speed planning scheme that is not dominated by any other solution is regarded as a non-dominated solution and saved to an external file. S2325. Randomly select a vehicle speed planning scheme from the external file as the global optimal solution.

6. The vehicle speed planning method based on cloud dynamic optimization according to claim 4, characterized in that The specific steps of step S233 are as follows: S2331. Construct the convergence index of the optimal solution during the population iteration process through the following formula: where \(C(t)\) is the convergence index of the \(t\)-th generation, \(m = 1, 2\) is the number of objectives, \(A(t)\) is the non-dominated solution set in the external archive, and \(|A(t)|\) is the number of non-dominated solutions in the external archive. is the \(m\)-th objective value of the \(n\)-th non-dominated solution in the external archive during the \((t - 1)\)-th iteration process; S2332. Construct the diversity index of the optimal solution during the population iteration process through the following formula: Among them, D ( t ) is the diversity index of the t th generation, S ( t ) is the distribution performance of the external archive at the t th iteration, d n ( t ) is the crowding distance of the n th non-dominated solution in the objective space in the external archive, is the average value of the crowding distances of all non-dominated solutions in the external archive; S2333. Set the exploitation performance to represent the performance of the population evolving by searching for new vehicle speed planning schemes based on the existing vehicle speed planning schemes, and set the exploration performance to represent the performance of the population evolving by optimizing the vehicle speeds of some road sections of the existing vehicle speed planning schemes. S2334. Judge the evolution state of the population. When C ( t ) > 0 and D ( t ) > 0, it is determined that both the exploitation performance and exploration performance of the population meet the requirements, and the current population evolution method is maintained; When C ( t ) ≥ 0 and D ( t ) < 0, it is determined that the exploration performance of the population is low, and the particle itself is further guided to evolve; When C ( t ) = 0 and D ( t ) > 0, it is determined that the development performance of the population is low, and the scope of population evolution is further expanded; S2335. Design Scaling Factor , and set the exploration performance of the population is enhanced when it increases, and set the exploitation performance of the population is enhanced when it decreases.

7. The vehicle speed planning method based on cloud dynamic optimization according to claim 4, wherein The specific steps of step S234 are as follows: S2341. Set the knowledge obtained during the population iteration process K ( T ) Construct the historical information of the individual optimal solution P i,d ( T ); Among them, K ( T ) is the knowledge obtained by the population in the t -th iteration process, P i,d ( T ) is the historical information composed of the optimal solutions of the i -th individual at different times, T = t - n +1, t - n +2, …, t is the matrix composed of historical iteration times, n is the preset parameter; S2342. Construct the cost function of the individual optimal trajectory prediction model based on knowledge reuse. Among them, is the parameter matrix, k is the number of samples, is the regression equation, expressed as follows: Among them, is the designed scaling factor, which is designed based on the adaptive dynamic feedback mechanism; S2343. Minimize the cost function and solve for the parameter matrix using the least squares method: ； Further derivation gives: ; Further simplification yields: Let , and the solution can be obtained as follows: ; S2344. Predict the individual optimal solution in the next iteration process according to the obtained parameter matrix : ； S2345. Determine whether the following is satisfied dominate ; If so, go to step S2346; If not, go to step S2347; S2346. Take as t the personal best solution of the i +1 generation and the first particle, and go to step S2348; S2347. Settings = As t The individual optimal solution of the i nth particle in the +1 generation; S2348. Randomly select a vehicle speed planning scheme from the external file as the global optimal solution; In step S235, update the velocity and position of the particle according to the individual optimal solution and the global optimal solution using the following formula: where v i,d ( t ) and x i,d ( t ) represent the velocity and position of the t -th iteration of the i -th particle in the d -th dimension, i = 1, 2, … N is the number of particles; d = 1, 2, …, D , p i,d ( t ) is the individual optimal solution of the i-th particle, g d ( t ) is the global optimal solution found by the population, w is the inertia weight, c 1 and c 2 are the individual cognitive factor and the social cognitive factor respectively, r 1 and r 2 are random numbers in the range [0, 1].

8. The vehicle speed planning method based on cloud dynamic optimization according to claim 4, characterized in that, The specific steps of step S236 are as follows: S2361. Judge whether the individual optimal solution is dominated by the vehicle speed planning schemes in the external file; If so, discard the individual optimal solution and go to step S237; If not, store the individual optimal solution in the external file and delete the vehicle speed planning schemes in the external file that are dominated by the individual optimal solution; S2362. Judge whether the number P of the vehicle speed planning schemes as non-dominated solutions in the external file is greater than the maximum size N of the external file; If so, go to step S2363; If not, go to step S237; S2363. Sort the non-dominated solutions in the external file according to the crowding distance, delete the non-dominated solution with the smallest crowding distance, and return to step S2362.

9. The vehicle speed planning method based on cloud dynamic optimization according to claim 3, wherein The specific steps of step S3 are as follows: S31. The cloud server associates each planned vehicle speed in the best planned vehicle speed vector with the corresponding road section; S32. The cloud server identifies the real-time road section to which the vehicle belongs and returns the planned vehicle speed corresponding to the real-time road section to the vehicle to guide the driver's driving; S33. The cloud server identifies whether the vehicle enters a new road section; If so, re-perform vehicle speed planning and return to step S1; If not, go to step S34; S34. Wait for a set time period and return to step S33.

10. A vehicle speed planning system based on cloud dynamic optimization, characterized in that, It includes: A data acquisition and road section division module, which is used to obtain the current state of the vehicle and the road information in front of the vehicle on the cloud server, and divide the road in front of the vehicle into road sections; A vehicle speed planning module, which is used to construct a multi-objective optimization model with the shortest driving time and the least number of stops of the vehicle on the front road as the optimization objectives and the vehicle speed of each road section as the decision variables on the cloud server, and use the multi-objective particle swarm optimization algorithm to solve the planned vehicle speed of each road section as the optimal solution. The planned speed feedback module is used to return the planned speed of the real-time section to which the vehicle belongs from the cloud server to the vehicle to guide the driver's driving.

Images