Multi-lane self-adaptive energy-saving cruise control method
Through the multi-lane adaptive energy-saving cruise control method, the self-learning Markov speed prediction and alternating direction multiplier method are used to optimize vehicle control, and the safety, comfort and energy-saving problems of vehicle cruise control in multi-lane environments are solved. Efficient traffic and energy consumption reduction are achieved.
Patent Information
- Application Number
- CN202411477368.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-22
- Publication Date
- 2024-11-22
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing technology is difficult to fully utilize traffic flow information for optimization and control in a single-lane environment, and cannot effectively solve the safety, comfort and energy-saving problems of vehicle cruise control in a multi-lane environment, especially the decision on horizontal lane change and vertical vehicle speed planning. Comprehensive difficulties.
Adopting multi-lane adaptive energy-saving cruise control method, road data and traffic flow information are obtained through V2X equipment, combined with self-learning Markov speed prediction and alternating direction multiplier method, optimize the longitudinal and lateral movements of the vehicle to achieve the lowest energy consumption Control policies and dynamically update lane selections to adapt to traffic environments.
It achieves the consideration of comfort, safety and energy saving effects in a multi-lane environment, improves traffic efficiency, has engineering application value, and shows high prediction accuracy and calculation efficiency through simulation verification.
Smart Images

Figure CN118991760A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of thermal management of electric vehicle, and in particular to a multi-lane adaptive energy-saving cruise control method. Background Art
[0002] With the application of advanced on-board sensors and the development of vehicle-road-network technology, ground vehicles can obtain information about the vehicle in front and the road ahead, which can be used to design predictive motion control, realize advanced assisted driving functions and even high-level unmanned driving technology. Among them, adaptive cruise control is an important technology for realizing high-level autonomous driving, and its control effect directly affects the safety, comfort and economy of vehicle following.
[0003] Therefore, adaptive cruise control methods based on road information and traffic conditions are of great significance to achieving safe, energy-saving and efficient assisted driving.
[0004] At present, vehicle cruise control technology has been widely studied. Existing methods are mainly carried out in a single-lane environment, ignoring the dynamics of traffic flow and the complexity of multiple lanes in real scenarios, and cannot fully utilize traffic flow information for optimization control. At present, there is a lack of multi-lane energy-saving cruise control schemes that consider both lateral lane change decisions and longitudinal vehicle speed planning. The trade-off between the uncertainty of road traffic flow, cruise control performance and computational load is the main difficulty in designing multi-lane adaptive energy-saving cruise control. Summary of the invention
[0005] The purpose of the present invention is to provide a multi-lane adaptive energy-saving cruise control method. The proposed multi-lane adaptive energy-saving cruise control method takes into account comfort, safety, energy-saving effect and traffic efficiency, and has engineering application value.
[0006] The present invention provides a multi-lane adaptive energy-saving cruise control method, comprising the following steps: S1, obtain the road data ahead and the surrounding traffic flow information, and sample the road slope in the spatial domain; Acquisition of road data ahead: The vehicle connects to the global geographic information system through the V2X device to obtain the altitude data of its driving route; Acquisition of surrounding traffic flow information: The vehicle detects the front and following vehicles in all lanes based on the on-board sensors and can measure the distance between vehicles.
[0007] Sampling the road slope in the spatial domain: discretize the road ahead with an equal interval ∆s of 10m, and sample the road slope and speed limit information.
[0008] S2, using a self-learning Markov speed prediction method to make a short-term prediction of the speed of the preceding vehicle; Offline training of the acceleration state transfer matrix: Before implementing Markov speed prediction, the acceleration state transfer matrix is trained using an offline database consisting of multiple standard driving conditions; S3, multi-lane energy-saving cruise optimization problem is established; Select optimization variables: Select the vehicle's nominal kinetic energy , travel time As the state variable, select the vehicle acceleration As the control quantity; using the longitudinal dynamics of the vehicle, the state transfer equation between kinetic energy, time and acceleration is established; ; ; in, is the time step; is the distance sampling interval; Indicates that at time step The nominal kinetic energy of the vehicle at ; Indicates different vehicles; Establish system variable constraints: The vehicle kinetic energy cannot exceed the upper limit corresponding to the road speed limit. To ensure comfort, the vehicle acceleration must be kept within a certain range. To ensure vehicle traffic efficiency, the minimum road speed is set. To limit the negative impact on following vehicles, the time interval constraint is used to filter the lane change space. Establish system variable constraint equations; ; ; ; is at the time step time; The vehicle's motor torque; It is the coefficient in the motor torque model; it is used to define the upper and lower limits of the motor torque. The face function represents the process of calculating the acceleration based on the motor torque and the maximum kinetic energy. Establish following vehicle safety constraints: The vehicle needs to meet following vehicle safety requirements, which constrains the safe following distance; ; Indicates at location The travel time of a vehicle at a certain location; Indicates the travel time of the preceding vehicle at the same position; Indicates the minimum time interval of operation; Design optimization goal: Minimum energy consumption is the optimization goal; ; in, Indicates the energy consumption per unit distance; Indicates the total number of sampling points; is a security-related penalty factor; it is used to weigh the importance of security in the optimization objective. Represents a safety redundant variable; S4, solving multi-lane energy-saving cruise optimization problem and online vehicle speed control; Determine the prediction domain, call the alternating direction multiplier method to calculate the optimization problem, use the optimal acceleration signal and lane change signal to control the lateral and longitudinal movements of the vehicle, and optimize and control the rolling forward.
[0009] Preferably, count the state transition events: count each current acceleration value Transfer to the corresponding next moment acceleration value The number of samples is denoted as , is the state space size; An index for the immediate future; ; is the acceleration grid length is the acceleration curve test length, and the offline trajectory is counted. The number of samples with equal values is denoted by ; State transition matrix calculation: According to the number of samples of state transition events, within the prediction time range, at a certain moment in the future q The transition probability matrix It can be obtained by the ratio of sample size; ; According to the sample size , , and the two are respectively related to the acceleration curve test length The ratio of represents the frequency of the two events in the curve, namely: ; Build frequency matrix , , then the state transfer matrix at a certain moment in the future can also be calculated as: ; Finally, according to the acceleration , , record its position index in the acceleration state grid , , storing the data at the corresponding position of the state transfer matrix.
[0010] Preferably, in step S2, a self-learning factor is defined: , , deriving its recursive form, we can get: ; in, Indicates that offline data is After the time, the acceleration changes from Transfer to the acceleration value The number of samples, , similarly, ; ; According to the recursive form of the two equations, the state transfer matrix at a certain moment can be calculated as: ; in Indicates the current acceleration in offline data from Transfer to its corresponding The acceleration value after time is The number of samples, ∈{0,1}, where is the introduced Markov learning coefficient; Online speed prediction: Get the speed and acceleration values at the current moment, get the corresponding transition probability vector according to the state transfer matrix, and get the acceleration prediction value ,Right now: ; Then, by placing the The predicted acceleration values are added together, and then the current speed value is added to get the The speed prediction value in the time step ,Right now: .
[0011] Preferably, in step S3, the vehicle energy consumption is calculated: the speed and torque of the drive motor are converted according to the vehicle speed and the output torque, the current efficiency of the motor is obtained by querying the motor characteristic diagram, and the current energy consumption of the vehicle is calculated in combination with the battery model.
[0012] Preferably, in step S3, the optimization problem is embedded in the model predictive control framework, and the distributed optimization algorithm is called to solve the model predictive control problem to obtain the optimal vehicle speed control strategy in the current prediction domain.
[0013] Preferably, in step S3, the lateral optimal lane change rule: the proposed new lane change algorithm selects the optimal lane according to the minimization of the lane cost; first, the cost of the current lane is initialized and the cost of the left lane and the cost of the right lane , and update the lane cost according to the above steps, and then introduce a threshold and penalty items It is used to limit the frequency and aggressiveness of lane changes, and at the same time check whether there is a potential collision risk in the left and right lanes. If the cost of a lane on one side is lower than the cost of the current lane and it is safe to change lanes, then choose to change lanes to that side; if the cost of the current lane is already the lowest, or the costs of lanes on both sides do not meet the lane change conditions, then stay in the current lane; after selecting the lane change direction, perform the corresponding lane changing operation to ensure that a safe distance is maintained during the lane change process to avoid collisions with other vehicles; finally, during driving, continuously monitor the surrounding traffic conditions and lane costs, and dynamically update the lane selection strategy to adapt to the changing traffic environment.
[0014] Preferably, in step S4, a solution based on the alternating direction multiplier method is designed to solve the multi-lane energy-saving cruise optimization problem: select the main variable and the dual variable, design the augmented Lagrangian function and form a standardized alternating direction multiplier method optimization problem, and use Newton's method to update the main and dual variables until the convergence conditions are met; forward rolling optimization control: update the vehicle position and prediction information, repeat the rolling optimization problem solution and online vehicle speed control.
[0015] Therefore, the present invention adopts the above-mentioned multi-lane adaptive energy-saving cruise control method. The proposed multi-lane adaptive energy-saving cruise control method takes into account comfort, safety, energy-saving effect and traffic efficiency, and has engineering application value.
[0016] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS
[0017] Figure 1 A schematic flow chart of a multi-lane adaptive energy-saving cruise control method of the present invention; Figure 2 A schematic diagram of a smart connected vehicle cruising on a multi-lane sloped road surface according to a multi-lane adaptive energy-saving cruise control method of the present invention; Figure 3 A schematic diagram of the steps for establishing a multi-lane energy-saving cruise optimization problem of a multi-lane adaptive energy-saving cruise control method of the present invention; Figure 4 A schematic diagram showing comparison of results of three speed prediction methods of a multi-lane adaptive energy-saving cruise control method of the present invention; Figure 5 A schematic diagram showing speed result comparison of three solution methods of a multi-lane adaptive energy-saving cruise control method of the present invention; Figure 6 A traffic flow analysis schematic diagram of a cruise control method of a multi-lane adaptive energy-saving cruise control method of the present invention. DETAILED DESCRIPTION
[0018] The technical solution of the present invention is further described below through the accompanying drawings and embodiments.
[0019] Unless otherwise defined, technical or scientific terms used in the present invention shall have the common meanings understood by one having ordinary skills in the field to which the present invention belongs.
[0020] The words "first", "second" and similar terms used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. Words such as "include" or "comprises" and similar terms mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Words such as "connect" or "connected" and similar terms are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the object being described changes, the relative positional relationship may also change accordingly.
[0021] Embodiment 1 like Figure 1-Figure 6 As shown, a multi-lane adaptive energy-saving cruise control method of the present invention comprises the following steps: S1, obtain the road data ahead and the surrounding traffic flow information, and sample the road slope in the spatial domain; Acquisition of road data ahead: The vehicle connects to the global geographic information system through the V2X device to obtain the altitude data of its driving route; Acquisition of surrounding traffic flow information: The vehicle detects the front and following vehicles in all lanes based on the on-board sensors and can measure the distance between vehicles.
[0022] Sampling the road slope in the spatial domain: discretize the road ahead with an equal interval ∆s of 10m, and sample the road slope and speed limit information.
[0023] S2, using a self-learning Markov speed prediction method to make a short-term prediction of the speed of the preceding vehicle; Offline training of the acceleration state transfer matrix: Before implementing Markov speed prediction, the acceleration state transfer matrix is trained using an offline database consisting of multiple standard driving conditions; In step S2, count the state transition events: count each current acceleration value Transfer to the corresponding next moment acceleration value The number of samples is denoted as , is the state space size; An index for the immediate future; ; is the acceleration grid length is the acceleration curve test length, and the offline trajectory is counted. The number of samples with equal values is denoted by ; State transition matrix calculation: According to the number of samples of state transition events, within the prediction time range, at a certain moment in the future q The transition probability matrix It can be obtained by the ratio of sample size; ; According to the sample size , , and the two are respectively related to the acceleration curve test length The ratio of represents the frequency of the two events in the curve, namely: ; Build frequency matrix , , then the state transfer matrix at a certain moment in the future can also be calculated as: ; Finally, according to the acceleration , , record its position index in the acceleration state grid , , storing the data at the corresponding position of the state transfer matrix.
[0024] Self-learning Markov speed prediction: Based on the current acceleration of the preceding vehicle and the acceleration state transfer matrix, the self-learning Markov prediction method is used to estimate the acceleration and speed trajectory of the preceding vehicle in the future. In step S2, the self-learning factor is defined: according to the event frequency , , deriving its recursive form, we can get: ; in, Indicates that offline data is After the time, the acceleration changes from Transfer to the acceleration value The number of samples, , similarly, ; ; According to the recursive form of the two equations, the state transfer matrix at a certain moment can be calculated as: ; in Indicates the current acceleration in offline data from Transfer to its corresponding The acceleration value after time is The number of samples, ∈{0,1}, where is the introduced Markov learning coefficient; Online speed prediction: Get the speed and acceleration values at the current moment, get the corresponding transition probability vector according to the state transfer matrix, and get the acceleration prediction value ,Right now: Then, by placing the The predicted acceleration values are added together, and then the current speed value is added to get the The speed prediction value in the time step ,Right now: S3, multi-lane energy-saving cruise optimization problem is established; Select optimization variables: Select the vehicle's nominal kinetic energy , travel time As the state variable, select the vehicle acceleration As the control quantity; using the longitudinal dynamics of the vehicle, the state transfer equation between kinetic energy, time and acceleration is established; in, is the time step; is the distance sampling interval; Indicates that at time step The nominal kinetic energy of the vehicle at ; Indicates different vehicles; Establish system variable constraints: The vehicle kinetic energy cannot exceed the upper limit corresponding to the road speed limit. To ensure comfort, the vehicle acceleration must be kept within a certain range. To ensure vehicle traffic efficiency, the minimum road speed is set. To limit the negative impact on following vehicles, the time interval constraint is used to filter the lane change space. Establish system variable constraint equations; ; ; ; is at the time step time; The vehicle's motor torque; It is the coefficient in the motor torque model; it is used to define the upper and lower limits of the motor torque. The face function represents the process of calculating the acceleration based on the motor torque and the maximum kinetic energy. Establish following vehicle safety constraints: The vehicle needs to meet following vehicle safety requirements, which constrains the safe following distance; ; Indicates at location The travel time of a vehicle at a certain location; Indicates the travel time of the preceding vehicle at the same position; Indicates the minimum time interval of operation; Design optimization goal: Minimum energy consumption is the optimization goal; ; in, Indicates the energy consumption per unit distance; Indicates the total number of sampling points; is a security-related penalty factor; it is used to weigh the importance of security in the optimization objective. Represents a safety redundant variable; In step S3, the vehicle energy consumption is calculated: the speed and torque of the drive motor are converted according to the vehicle speed and the output torque, the current efficiency of the motor is obtained by querying the motor characteristic diagram, and the current energy consumption of the vehicle is calculated in combination with the battery model.
[0025] Design a longitudinal vehicle speed model predictive control strategy and a lateral optimal lane-changing rule; with the goal of minimizing energy consumption, design a longitudinal vehicle speed model predictive control strategy and a lateral optimal lane-changing rule in the spatial domain.
[0026] In step S3, the optimization problem is embedded in the model predictive control framework, and the distributed optimization algorithm is called to solve the model predictive control problem to obtain the optimal vehicle speed control strategy in the current prediction domain.
[0027] In step S3, the lateral optimal lane-changing rule: The proposed new lane-changing algorithm selects the optimal lane according to the minimization of the lane cost; first, the cost of the current lane is initialized and the cost of the left lane and the cost of the right lane , and update the lane cost according to the above steps, and then introduce a threshold and penalty items It is used to limit the frequency and aggressiveness of lane changes, and at the same time check whether there is a potential collision risk in the left and right lanes. If the cost of a lane on one side is lower than the cost of the current lane and it is safe to change lanes, then choose to change lanes to that side; if the cost of the current lane is already the lowest, or the costs of lanes on both sides do not meet the lane change conditions, then stay in the current lane; after selecting the lane change direction, perform the corresponding lane changing operation to ensure that a safe distance is maintained during the lane change process to avoid collisions with other vehicles; finally, during driving, continuously monitor the surrounding traffic conditions and lane costs, and dynamically update the lane selection strategy to adapt to the changing traffic environment.
[0028] S4, solving multi-lane energy-saving cruise optimization problem and online vehicle speed control; Determine the prediction domain, call the alternating direction multiplier method to calculate the optimization problem, use the optimal acceleration signal and lane change signal to control the lateral and longitudinal movements of the vehicle, and optimize and control the rolling forward.
[0029] In step S4, a solution based on the alternating direction multiplier method is designed to solve the multi-lane energy-saving cruise optimization problem: the main variable and the dual variable are selected, the augmented Lagrangian function is designed and a standardized alternating direction multiplier method optimization problem is formed, and the Newton method is used to update the main and dual variables until the convergence conditions are met; forward rolling optimization control: the vehicle position and prediction information are updated, and the rolling optimization problem solution and online vehicle speed control are repeated.
[0030] To further verify the advantages of the present invention, a virtual environment was developed that can accurately reproduce real-world conditions, including road topography and surrounding traffic conditions, using open source maps to replicate an urban highway in Nanjing, which features a maximum speed limit of 80 km / h. Next, random traffic flows were generated that simulated the longitudinal and lateral movements of nearby vehicles.
[0031] The accuracy of the proposed self-learning Markov prediction method is evaluated by randomly selecting vehicle speed trajectories from the traffic flow. For comparative analysis, two other prediction methods are developed: a conventional Markov chain method and a neural network method. The effectiveness of the alternating direction multiplier method is also verified by comparative analysis with the discrete DP method and the widely adopted interior point method (IP).
[0032] like Figure 4 As shown in the figure, the speed prediction results of different methods: (a) Markov chain method, (b) LSTM method, and (c) self-learning Markov method. The metric for evaluating the prediction accuracy is the root mean square error (RMSE). Under the self-learning Markov method used, the average error is smaller and the performance is better.
[0033] like Figure 5As shown, the proposed alternating direction multiplier method exhibits comparable performance trends and response times to the DP and IP methods. A comprehensive summary of the specific performance metrics is shown in Table 2.
[0034] like Figure 6 Figure 2 shows the speeds of the ego vehicle and surrounding vehicles in the MLEC simulation. The ego vehicle can achieve a higher speed without causing the surrounding vehicles to slow down.
[0035] Table 1 is a performance comparison of the speed prediction method with the conventional Markov chain method and the neural network method.
[0036] Table 1
[0037] The RMSE of Markov, LSTM and self-learning Markov methods are 1.08 m / s, 0.91 m / s and 0.87 m / s respectively, which shows that the accuracy of LSTM and self-learning Markov methods is improved by 15.74% and 19.44% compared with the traditional Markov method. In addition, the experimental results of self-learning Markov show that the computational efficiency of this method is 0.6ms, which is significantly faster than the LSTM method and nearly 40 times faster.
[0038] Table 2 is a performance comparison of three optimization solution methods, among which ADMM is the alternating direction multiplier method.
[0039] Table 2
[0040] In terms of optimization cost, the DP method achieves the minimum cost of 77.23, while IP and ADMM record slightly higher costs of 78.68 and 78.65, respectively. It is worth noting that the optimality deviations of IP and ADMM are small, about 2% and 1.5%. In addition, the computation time is also a metric to evaluate the performance, and as pointed out in Table 2, DP requires significantly more computation time. In contrast, the alternating direction multiplier method has the shortest computation time, which only takes 5 milliseconds.
[0041] Table 3 is a performance comparison of five cruise controllers, among which IDM-Onelane is a single-lane intelligent driver controller, EC-Onelane is a single-lane energy-saving cruise control, and EC-MOBIL is a multi-lane energy-saving cruise control based on human lane-changing intention.
[0042] Table 3
[0043] Comparing MLEC with EC-MOBIL, MLEC can save 6.68% of energy, which is 2.99% higher than EC-MOBIL, indicating that the proposed lane-changing strategy can reduce energy consumption by 2.99%; compared with GOEC, MLEC's proposed energy saving effect is slightly lower by 0.94%, while its time saving is slightly higher than GOEC by 1.48%. In short, the performance of MLEC is comparable to that of GOEC.
[0044] The self-learning Markov speed prediction method outperforms the traditional Markov chain method and LSTM method in terms of accuracy; in addition, the newly developed alternating direction multiplier method provides near-optimal speed planning performance, comparable to the DP solution, and significantly improves computational efficiency. The MLEC control proposed based on micro-dynamic traffic simulation combined with real-world terrain is superior to existing cruise control that only considers speed optimization or uses humanoid lane change strategies. The results of MLEC are comparable to those of the optimal controller GOEC, with energy consumption and driving time reduced by 6.68% and 24.56%, respectively. In addition, the results of the HIL test confirmed the performance consistency of MLEC with the simulation, revealing an average computation time of 23.5ms on a real-time control processing system, thus verifying its ability for onboard applications. Therefore, the invented method has good potential for online applications.
[0045] Therefore, the present invention adopts the above-mentioned multi-lane adaptive energy-saving cruise control method. The proposed multi-lane adaptive energy-saving cruise control method takes into account comfort, safety, energy-saving effect and traffic efficiency, and has engineering application value.
[0046] The above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that they can still modify or replace the technical solution of the present invention with equivalents, and these modifications or equivalent replacements cannot cause the modified technical solution to deviate from the spirit and scope of the technical solution of the present invention.
Claims
1. A multi-lane adaptive energy-saving cruise control method, characterized in that: The following steps are involved: S1, obtain the road data ahead and the surrounding traffic flow information, and sample the road slope in the spatial domain; Acquisition of road data ahead: The vehicle connects to the global geographic information system through the V2X device to obtain the altitude data of its driving route; Acquisition of surrounding traffic flow information: The vehicle detects the front and following vehicles in all lanes based on the on-board sensors and can measure the distance between vehicles. Sampling the road slope in the spatial domain: discretize the road ahead with an equal interval ∆s of 10m, and sample the road slope and speed limit information. S2, using a self-learning Markov speed prediction method to make a short-term prediction of the speed of the preceding vehicle; Offline training of the acceleration state transfer matrix: Before implementing Markov speed prediction, the acceleration state transfer matrix is trained using an offline database consisting of multiple standard driving conditions; S3, multi-lane energy-saving cruise optimization problem is established; Select optimization variables: Select the vehicle's nominal kinetic energy , travel time As the state variable, select the vehicle acceleration As the control quantity; using the longitudinal dynamics of the vehicle, the state transfer equation between kinetic energy, time and acceleration is established; ; ; in, is the time step; is the distance sampling interval; Indicates that at time step The nominal kinetic energy of the vehicle at ; Indicates different vehicles; Establish system variable constraints: The vehicle kinetic energy cannot exceed the upper limit corresponding to the road speed limit. To ensure comfort, the vehicle acceleration must be kept within a certain range. To ensure vehicle traffic efficiency, the minimum road speed is set. To limit the negative impact on following vehicles, the time interval constraint is used to filter the lane change space. Establish system variable constraint equations; ; ; ; is at the time step time; The vehicle's motor torque; It is the coefficient in the motor torque model; it is used to define the upper and lower limits of the motor torque. The face function represents the process of calculating the acceleration based on the motor torque and the maximum kinetic energy. Establish following vehicle safety constraints: The vehicle needs to meet following vehicle safety requirements, which constrains the safe following distance; ; Indicates at location The travel time of a vehicle at a certain location; Indicates the travel time of the preceding vehicle at the same position; Indicates the minimum time interval of operation; Design optimization goal: Minimum energy consumption is the optimization goal; ; in, Indicates the energy consumption per unit distance; Indicates the total number of sampling points; is a security-related penalty factor; it is used to weigh the importance of security in the optimization objective. Represents a safety redundant variable; S4, solving multi-lane energy-saving cruise optimization problem and online vehicle speed control; Determine the prediction domain, call the alternating direction multiplier method to calculate the optimization problem, use the optimal acceleration signal and lane change signal to control the lateral and longitudinal movements of the vehicle, and optimize and control the rolling forward.
2. A multi-lane adaptive energy-saving cruise control method according to claim 1, characterized in that: In step S2, count the state transition events: count each current acceleration value Transfer to the corresponding next moment acceleration value The number of samples is denoted as , is the state space size; An index for the immediate future; ; is the acceleration grid length is the acceleration curve test length, and the offline trajectory is counted. The number of samples with equal values is denoted by ; State transition matrix calculation: According to the number of samples of state transition events, within the prediction time range, at a certain moment in the future q The transition probability matrix It can be obtained by the ratio of sample size; ; According to the sample size , , and the two are respectively related to the acceleration curve test length The ratio of represents the frequency of the two events in the curve, namely: ; Build frequency matrix , , then the state transfer matrix at a certain moment in the future can also be calculated as: ; Finally, according to the acceleration , , record its position index in the acceleration state grid , , storing the data at the corresponding position of the state transfer matrix.
3. The multi-lane adaptive energy-saving cruise control method according to claim 1, characterized in that: In step S2, the self-learning factor is defined: according to the event frequency , , deriving its recursive form, we can get: ; in, Indicates that offline data is After the time, the acceleration changes from Transfer to the acceleration value The number of samples, , similarly, ; ; According to the recursive form of the two equations, the state transfer matrix at a certain moment can be calculated as: ; in Indicates the current acceleration in offline data from Transfer to its corresponding The acceleration value after time is The number of samples, ∈{0,1}, where is the introduced Markov learning coefficient; Online speed prediction: Get the speed and acceleration values at the current moment, get the corresponding transition probability vector according to the state transfer matrix, and get the acceleration prediction value ,Right now: ; Then, by placing the The predicted acceleration values are added together, and then the current speed value is added to get the The speed prediction value in the time step ,Right now: 。 4. The multi-lane adaptive energy-saving cruise control method according to claim 1, characterized in that: In step S3, the vehicle energy consumption is calculated: the speed and torque of the drive motor are converted according to the vehicle speed and the output torque, the current efficiency of the motor is obtained by querying the motor characteristic diagram, and the current energy consumption of the vehicle is calculated in combination with the battery model.
5. The multi-lane adaptive energy-saving cruise control method according to claim 1, characterized in that: In step S3, the optimization problem is embedded in the model predictive control framework, and the distributed optimization algorithm is called to solve the model predictive control problem to obtain the optimal vehicle speed control strategy in the current prediction domain.
6. The multi-lane adaptive energy-saving cruise control method according to claim 1, characterized in that: In step S3, the lateral optimal lane-changing rule: The proposed new lane-changing algorithm selects the optimal lane according to the minimization of the lane cost; first, the cost of the current lane is initialized and the cost of the left lane and the cost of the right lane , and update the lane cost according to the above steps, and then introduce a threshold and penalty items It is used to limit the frequency and aggressiveness of lane changes, and at the same time check whether there is a potential collision risk in the left and right lanes. If the cost of a lane on one side is lower than the cost of the current lane and it is safe to change lanes, then choose to change lanes to that side; if the cost of the current lane is already the lowest, or the costs of lanes on both sides do not meet the lane change conditions, then stay in the current lane; after selecting the lane change direction, perform the corresponding lane changing operation to ensure that a safe distance is maintained during the lane change process to avoid collisions with other vehicles; finally, during driving, continuously monitor the surrounding traffic conditions and lane costs, and dynamically update the lane selection strategy to adapt to the changing traffic environment.
7. The multi-lane adaptive energy-saving cruise control method according to claim 1, characterized in that: In step S4, a solution based on the alternating direction multiplier method is designed to solve the multi-lane energy-saving cruise optimization problem: the main variable and the dual variable are selected, the augmented Lagrangian function is designed and the standardized alternating direction multiplier method optimization problem is formed, and the Newton method is used to update the main and dual variables until the convergence conditions are met; Forward rolling optimization control: Update vehicle position and prediction information, repeat rolling optimization problem solving and online vehicle speed control.
