Brake energy recovery method, device and vehicle

Abstract

The embodiment of the application relates to the technical field of vehicles, and discloses a brake energy recovery method and device and a vehicle, the method comprising the following steps: constructing a risk potential diagram corresponding to a future driving path of a vehicle based on scene sensing information of the vehicle; generating a driving suggestion of the future driving path based on the scene sensing information and driving state information of the vehicle; optimizing and solving a multi-target optimization function according to the risk potential diagram and the driving suggestion, obtaining an optimal braking curve corresponding to the future driving path, and controlling the vehicle to recover brake energy by using the optimal braking curve; wherein the driving suggestion conforms to a driving habit of a driver of the vehicle in terms of driving speed, and multiple optimization targets of the multi-target optimization function comprise that a total risk potential of the future driving path is minimum, brake energy recovery efficiency of the future driving path is maximum, and a gap between the braking curve of the future driving path and the driving suggestion is minimum. The technical scheme provided by the embodiment of the application improves the driving experience of a user.

Description

Technical Field

[0001] This application relates to the field of vehicle technology, specifically to a braking energy recovery method, device, and vehicle. Background Technology

[0002] Regenerative braking refers to the process of converting wasted kinetic energy into electrical energy and storing it during vehicle deceleration or braking.

[0003] In related technologies, during the braking energy recovery process, the requirements of the vehicle energy recovery system are determined according to the vehicle type and application scenario. Based on the vehicle's dynamic model and the requirements of the vehicle energy recovery system, an energy recovery strategy is formulated using intelligent optimization algorithms. Then, based on the energy recovery strategy and the requirements of the vehicle energy recovery system, an energy recovery system is designed so that the energy recovery system can adapt to braking energy recovery under various driving conditions and improve energy utilization.

[0004] However, the energy recovery strategies provided by the above-mentioned technical solutions cannot meet the actual needs and driving habits of drivers. When automatic braking is performed according to the energy recovery strategy, it is easy to conflict with the driver's intentions and driving habits, and even produce a dragging feeling and a jerking feeling, which reduces the user's driving experience. Summary of the Invention

[0005] In view of the shortcomings of the prior art, the purpose of this application is to provide a braking energy recovery method, device and vehicle that improves the driver's user experience.

[0006] In a first aspect, embodiments of this application provide a braking energy recovery method, the method comprising: constructing a risk potential energy map corresponding to the future driving path of the vehicle based on the vehicle's scene perception information; generating driving suggestions for the future driving path based on the scene perception information and the vehicle's driving state information; optimizing and solving a multi-objective optimization function according to the risk potential energy map and the driving suggestions to obtain the optimal braking curve corresponding to the future driving path; and using the optimal braking curve to control the vehicle to recover braking energy.

[0007] The risk potential energy map characterizes the relationship between the driving speed at a point on the future driving path and the risk potential energy of the driving scenario at that point. The driving recommendations align with the driver's driving habits at the given speed. Multiple optimization objectives in the multi-objective optimization function include: minimizing the total risk potential energy of the future driving path, maximizing the braking energy recovery efficiency of the future driving path, and minimizing the difference between the braking curve of the future driving path and the driving recommendations. The optimal braking curve is used to control the vehicle for braking energy recovery.

[0008] The technical solution provided in this application generates driving suggestions that conform to the driver's driving habits by using scene perception information and vehicle driving status information. Based on the driving suggestions, a multi-objective optimization function is used to determine the optimal braking curve. When regenerating braking energy according to the optimal braking curve, the vehicle's speed, deceleration, and deceleration changes are more in line with the driver's driving habits, avoiding dragging and jerking sensations during braking energy recovery, achieving a sense of unity between the driver and the vehicle, and improving the user's driving experience. At the same time, by comprehensively considering the risk potential energy in complex scenarios, the driver's driving habits, and the efficiency of braking energy recovery through the multi-objective optimization function, decision conflicts in complex scenarios are avoided, improving the safety of intelligent driving control, and the vehicle's range is improved by maximizing energy utilization.

[0009] One possible implementation involves constructing a risk potential energy map corresponding to the vehicle's future driving path based on the vehicle's scene perception information. Specifically, this can be achieved by: determining the driving scenario of the target location based on the scene perception information; identifying the target risk type associated with the driving scenario; constructing a correspondence between the driving speed of the target location and the risk potential energy of the driving scenario based on the target risk type; and generating a risk potential energy map based on this correspondence. Here, the driving scenario can be a single traffic scenario or a composite scenario of multiple traffic scenarios; the target location is each location point on the future driving path; and the target risk type is the risk type in the risk type dataset aggregated from the vehicle's driving. Different driving scenarios correspond to different risk potential energies, accurately describing the risk situation of different driving scenarios and improving the accuracy and relevance of risk decision-making.

[0010] One possible implementation involves traffic scenarios including: curve scenarios, speed bump scenarios, traffic light scenarios, congestion scenarios, and ramp scenarios. The risk type dataset includes: forward collision risk types, curve loss of control risk types, speed bump discomfort risk types, traffic light violation risk types, and speed limit violation risk types. By using multiple traffic scenarios and risk types, and refining the simulation of real-world vehicle driving scenarios and risks, more accurate and comprehensive braking energy recovery strategies can be developed for complex situations, improving the flexibility, safety, and reliability of intelligent driving.

[0011] One possible implementation involves having multiple target risk types. Constructing a correspondence between the driving speed of a target location and the risk potential energy of the driving scenario in which the target location is located, based on the target risk types, can be specifically implemented as follows: Based on the first mapping relationship and weight coefficients corresponding to each of the multiple target risk types, a correspondence is constructed between the driving speed of the target location and the risk potential energy of the driving scenario in which the target location is located. The first mapping relationship characterizes the correspondence between a location point, its driving speed, and its risk potential energy under the corresponding target risk type. Each risk type has its own weight coefficient, which allows for flexible adjustment of the importance of each risk type in the risk potential energy map. This enables greater reference to risk types with higher importance when determining the optimal braking curve, thus improving the accuracy of the optimal braking curve.

[0012] One possible implementation involves determining the weight coefficients for multiple target risk types, including: determining a target style adjustment factor based on the driver's driving style and a dataset showing the first mapping relationship between driving style and style adjustment factors; determining a target environment adjustment factor based on the vehicle's environmental state and a dataset showing the second mapping relationship between environmental state and environment adjustment factors; and determining the weight coefficients for each of the multiple target risk types based on the target style adjustment factor, the target environment adjustment factor, and the basic weight coefficients corresponding to each target risk type. Driving style includes aggressive and conservative types, and environmental state includes road adhesion state and visibility state. Real-time perception of driving style and environmental state, and automatic adjustment of weight coefficients based on these factors, offers greater flexibility and results in a more accurate and realistic risk potential map.

[0013] One possible implementation involves determining the weighting coefficients for each of the multiple target risk types based on the target style adjustment factor, the target environment adjustment factor, and the basic weighting coefficients corresponding to each target risk type. Specifically, this can be achieved by: determining the adjustment ratio as the product of the target style adjustment factor and the target environment adjustment factor; determining the adjustment ratio limit for each of the multiple target risk types based on their respective risk levels; and determining the corresponding weighting coefficients for each of the multiple target risk types based on the adjustment ratio, the adjustment ratio limits for each target risk type, and the basic weighting coefficients. By combining driving style and environmental conditions to adjust the weighting coefficients, and comprehensively considering the factors affecting the weighting coefficient changes from multiple dimensions, the adjustment results are more accurate.

[0014] One possible implementation involves driving recommendations that include a target speed at the end of the future driving path and a desired braking deceleration curve for that path. The target speed and desired braking deceleration are specific driving parameter suggestions, which, compared to driving style labels, provide a more refined reflection of the driver's driving habits.

[0015] One possible implementation involves using driving status information, including the driving speed and braking deceleration at each moment within the current acquisition time window. The current acquisition time window is a time window consisting of the current moment and several moments preceding it. Generating driving suggestions for future routes based on scene perception information and driving status information can be specifically implemented as follows: determining the driving scenario of the target location based on scene perception information; and generating driving suggestions for future routes based on the driving scenario of the target location, driving status information, and a driving habit learning model. Here, the driving scenario can be a single traffic scenario or a composite scenario of multiple traffic scenarios; the target location is each location point on the future driving route; and the driving habit learning model is trained using historical sample data, which includes: the driving speed and braking deceleration at each moment within the acquisition time window corresponding to a historical moment, the driving scenario of each location point on the future driving route corresponding to a historical moment, and the actual braking curve of the future driving route corresponding to a historical moment. By outputting driving suggestions for future routes that conform to the driver's driving habits through the driving habit learning model, the process of determining the braking curve can be guided more efficiently, intelligently, and predictively, providing a reference target for the braking curve and improving the efficiency of braking curve determination.

[0016] One possible implementation involves using a braking curve along the future driving path to reflect the driving speed and braking deceleration at each point on the future driving path. The difference between the braking curve and the driving suggestion is determined based on a first deviation value and a second deviation value. The first deviation value is the sum of the deviations between the driving speed and the target speed at each point on the future driving path, and the second deviation value is the sum of the deviations between the braking deceleration and the desired braking deceleration at each point on the future driving path. The desired braking deceleration at each point on the future driving path is obtained by querying the desired braking deceleration curve. The construction process of the multi-objective optimization function includes: constructing a first optimization term with the objective of minimizing the total risk potential energy of the future driving path; constructing a second optimization term with the objective of maximizing braking energy recovery efficiency; constructing a third optimization term with the minimum first deviation value; and constructing a fourth optimization term with the minimum second deviation value. Based on the first, second, third, and fourth optimization terms, as well as their respective optimization weights and constraints, the multi-objective optimization function is constructed. The braking energy recovery efficiency is determined based on the braking deceleration at each point along the future driving path. Aiming to minimize the first deviation value, the second deviation value, and the total risk potential energy, it reduces the gap between the braking curve and the driving recommendation while lowering driving risk, maximizing alignment with the driver's driving habits. Simultaneously, it maximizes energy recovery efficiency, increasing the energy recovered during braking.

[0017] One possible implementation involves the driving suggestion including the desired deceleration curve shape. A multi-objective optimization function is constructed based on the first, second, third, and fourth optimization terms, along with their respective optimization weights and constraints. Specifically, this can be achieved by constructing a fifth optimization term that minimizes the rate of change of braking deceleration at each point along the future driving path. This multi-objective optimization function is built based on the first, second, third, fourth, and fifth optimization terms, along with their respective optimization weights and constraints. The optimization weights of the fifth optimization term are taken from the weight variation range corresponding to the desired deceleration curve shape. Adjusting the braking deceleration change by modifying the desired deceleration curve shape further improves the fit between the braking curve and the driving suggestion.

[0018] Secondly, this application provides a braking energy recovery device, which includes a processing module and a braking module.

[0019] The aforementioned processing module is used to construct a risk potential energy map corresponding to the vehicle's future driving path based on the vehicle's scene perception information.

[0020] The aforementioned processing module is also used to generate driving suggestions for future driving routes based on scene perception information and vehicle driving status information.

[0021] The aforementioned processing module is also used to optimize the multi-objective optimization function based on the risk potential energy map and driving suggestions, so as to obtain the optimal braking curve corresponding to the future driving path.

[0022] The aforementioned braking module is used to control the vehicle for brake energy recovery using the optimal braking curve.

[0023] Among them, the risk potential energy map is used to characterize the correspondence between the driving speed of a location point on the future driving path and the risk potential energy of the driving scenario in which the location point is located. The driving suggestions are in line with the driving habits of the driver of the vehicle at driving speed. The multiple optimization objectives in the multi-objective optimization function include: minimizing the total risk potential energy of the future driving path, maximizing the braking energy recovery efficiency of the future driving path, and minimizing the difference between the braking curve of the future driving path and the driving suggestions.

[0024] One possible implementation is that the aforementioned processing module is specifically used for: determining the driving scenario in which the target location point is located based on scene perception information; determining the target risk type associated with the driving scenario in which the target location point is located; constructing a correspondence between the driving speed of the target location point and the risk potential energy of the driving scenario in which the target location point is located based on the target risk type; and generating a risk potential energy map based on the correspondence between the driving speed of the target location point and the risk potential energy of the driving scenario in which the target location point is located. Here, the driving scenario is a single traffic scenario or a composite scenario of multiple traffic scenarios, the target location point is each location point on the future driving path, and the target risk type is the risk type in the risk type dataset aggregated from the vehicle's driving.

[0025] One possible implementation involves traffic scenarios including: curve scenarios, speed bump scenarios, traffic light scenarios, congestion scenarios, and ramp scenarios. The risk type dataset includes: forward collision risk type, curve loss of control risk type, speed bump inconvenience risk type, traffic light violation risk type, and speed limit violation risk type.

[0026] One possible implementation involves having multiple target risk types. Specifically, the aforementioned processing module is used to: construct a correspondence between the driving speed of a target location and the risk potential energy of the driving scenario in which the target location is located, based on the first mapping relationship and weight coefficients corresponding to each of the multiple target risk types. The first mapping relationship is used to characterize the correspondence between a location point, its driving speed, and its risk potential energy under the corresponding target risk type.

[0027] One possible implementation is that the aforementioned processing module is specifically used to: determine a target style adjustment factor based on the driver's driving style and a dataset of first mapping relationships between driving style and style adjustment factors; determine a target environment adjustment factor based on the vehicle's environmental state and a dataset of second mapping relationships between environmental state and environment adjustment factors; and determine the weight coefficients corresponding to multiple target risk types based on the target style adjustment factor, the target environment adjustment factor, and the basic weight coefficients corresponding to each of the multiple target risk types. Here, driving style includes aggressive and conservative, and environmental state includes road adhesion state and visibility state.

[0028] One possible implementation, the aforementioned processing module, is specifically used to: determine the adjustment ratio by multiplying the target style adjustment factor and the target environment adjustment factor; determine the adjustment ratio limits for each of the multiple target risk types based on their respective risk levels; and determine the corresponding weight coefficients for each of the multiple target risk types based on the adjustment ratio, the adjustment ratio limits for each of the multiple target risk types, and the basic weight coefficients.

[0029] One possible implementation involves driving recommendations that include: a target speed at the end point of the future driving path and a desired braking deceleration curve for the future driving path.

[0030] One possible implementation involves the driving status information including the driving speed and braking deceleration at each moment within the current acquisition time window. The current acquisition time window is a time window consisting of the current moment and multiple moments preceding it. Specifically, the processing module is used to: determine the driving scenario of the target location based on scene perception information; and generate driving suggestions for future routes based on the driving scenario of the target location, the driving status information, and the driving habit learning model. The driving scenario can be a single traffic scenario or a composite scenario of multiple traffic scenarios. The target location is each location point on the future driving route. The driving habit learning model is trained using historical sample data, which includes: the driving speed and braking deceleration at each moment within the acquisition time window corresponding to a historical moment, the driving scenario of each location point on the future driving route corresponding to a historical moment, and the actual braking curve of the future driving route corresponding to a historical moment.

[0031] One possible implementation involves using a braking curve along the future driving path to reflect the driving speed and braking deceleration at each point along the path. The difference between the braking curve and the driving suggestion is determined based on a first deviation value and a second deviation value. The first deviation value is the sum of the deviations between the driving speed and the target speed at each point along the future driving path, and the second deviation value is the sum of the deviations between the braking deceleration and the desired braking deceleration at each point along the future driving path. The desired braking deceleration at each point along the future driving path is obtained by querying the desired braking deceleration curve. Specifically, the processing module is used to: construct a first optimization term aiming to minimize the total risk potential energy of the future driving path; construct a second optimization term aiming to maximize braking energy recovery efficiency; construct a third optimization term aiming to minimize the first deviation value; and construct a fourth optimization term aiming to minimize the second deviation value. Based on the first, second, third, and fourth optimization terms, as well as their respective optimization weights and constraints, a multi-objective optimization function is constructed. The braking energy recovery efficiency is determined based on the braking deceleration at each point along the future driving path.

[0032] One possible implementation involves providing driving suggestions that include the desired deceleration curve shape. Specifically, the aforementioned processing module is used to: construct a fifth optimization term with the objective of minimizing the rate of change of braking deceleration at each location point along the future driving path; and to construct a multi-objective optimization function based on the first, second, third, fourth, and fifth optimization terms, as well as their respective optimization weights and constraints. The optimization weights of the fifth optimization term are taken from the weight change interval corresponding to the desired deceleration curve shape.

[0033] The technical effects of any implementation method in the second aspect can be found in the technical effects of any implementation method in the first aspect mentioned above, and will not be repeated here.

[0034] Thirdly, this application provides a vehicle that includes the brake energy recovery device in any of the embodiments of the second aspect above, or the vehicle uses the brake energy recovery method in any of the embodiments of the first aspect above to recover brake energy.

[0035] Fourthly, this application provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the braking energy recovery method in any embodiment of the first aspect described above.

[0036] Fifthly, this application provides a computer program product, which includes a computer program or instructions that, when executed by a processor, implement the braking energy recovery method in any of the embodiments of the first aspect described above.

[0037] The solutions provided in the third to fifth aspects above can realize the braking energy recovery method in any embodiment of the first aspect above, and their specific implementations will not be described in detail here. The technical effects corresponding to any implementation of the solutions provided in the third to fifth aspects above can be found in the technical effects corresponding to any implementation of the first aspect above, and will not be described in detail here.

[0038] It should be noted that any of the possible implementations of any of the above aspects can be combined, provided that the solutions do not contradict each other. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of this application or the background art, the accompanying drawings used in the embodiments of this application will be described below.

[0040] Figure 1 This is a schematic diagram of a braking energy recovery system provided in an embodiment of this application;

[0041] Figure 2 A schematic flowchart of a braking energy recovery method provided in an embodiment of this application;

[0042] Figure 3 This is a schematic diagram of a braking energy recovery architecture provided in an embodiment of this application;

[0043] Figure 4 A schematic diagram of module interaction for a braking energy recovery architecture provided in an embodiment of this application;

[0044] Figure 5 A schematic flowchart of another braking energy recovery method provided in an embodiment of this application;

[0045] Figure 6 A schematic diagram of a driving speed curve provided in an embodiment of this application;

[0046] Figure 7 A flowchart illustrating a deceleration curve optimization method provided in an embodiment of this application;

[0047] Figure 8 A schematic diagram of another driving speed curve provided in an embodiment of this application;

[0048] Figure 9 A schematic diagram of another driving speed curve provided in an embodiment of this application;

[0049] Figure 10 This is a schematic diagram of a braking energy recovery device provided in an embodiment of this application. Detailed Implementation

[0050] To enable those skilled in the art to better understand the technical solutions of this application, the technical solutions in the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings.

[0051] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0052] In the embodiments of this application, the words "exemplary," "for example," or "for instance" are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary," "for example," or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the words "exemplary," "for example," or "for instance" is intended to present the relevant concepts in a specific manner.

[0053] The embodiments of this application are described below with reference to the accompanying drawings.

[0054] This application provides a vehicle.

[0055] Alternatively, a vehicle may also be referred to as a vehicle, mobile carrier, electric vehicle (EV), hybrid electric vehicle (HEV), plug-in hybrid electric vehicle (PHEV), fuel cell vehicle (FCV), autonomous vehicle, intelligent and connected vehicle (ICV), driverless vehicle, or new energy vehicle. In this application embodiment, the vehicle may be a sedan, sport utility vehicle (SUV), truck, electric vehicle, motorcycle, tricycle, special vehicle (such as ambulance, fire truck, police car, etc.), driverless taxi, intelligent connected bus, autonomous logistics vehicle, electric truck, etc. The method provided in this application embodiment is also applicable to various special-purpose vehicles, such as agricultural vehicles, mining vehicles, forestry vehicles, airport vehicles, and port vehicles; this application does not impose specific limitations on these.

[0056] In some embodiments, the vehicle includes a brake energy recovery system.

[0057] Among them, regenerative braking refers to converting the kinetic energy (motion energy) that would otherwise be wasted into electrical energy and storing it when a vehicle decelerates or brakes.

[0058] For example, when a driver presses the brake pedal, the brake calipers clamp the brake discs, causing the wheels to slow down through friction. During deceleration, the vehicle's kinetic energy is converted into heat and dissipated into the air. Regenerative braking sends control commands to the vehicle's motor during braking, switching it from drive mode to generator mode. In this mode, the motor acts as a generator. The rotor, connected to the wheels, rotates in a magnetic field, generating electromagnetic resistance. This resistance acts on the wheels, slowing the vehicle down, while the wheels' kinetic energy is converted into electrical energy and stored in the vehicle's battery. Specifically, the motor controller inputs current into the motor's stator windings to generate a magnetic field. When the wheels drive the rotor to rotate in this magnetic field, they cut magnetic lines of field, inducing a current in the stator windings. This induced current interacts with the rotor's magnetic field, creating electromagnetic resistance in the opposite direction of rotation. This resistance acts on the wheels, braking the vehicle. The vehicle's kinetic energy overcomes this resistance, continuously and stably generating an induced current in the stator windings. This induced current is stored as recovered electrical energy in the vehicle's battery.

[0059] In related technologies, the requirements for vehicle energy recovery systems are determined based on vehicle type and application scenarios. Based on the vehicle's dynamics model and system requirements, intelligent optimization algorithms are used to formulate energy recovery strategies. Then, based on these strategies and system requirements, an energy recovery system is designed to adapt to braking energy recovery under various driving conditions, improving energy utilization. However, current braking energy recovery strategies only consider energy utilization efficiency and fail to align with the driver's actual needs and driving habits. This can easily conflict with the driver's intentions and habits during automatic braking and deceleration, reducing the user's driving experience.

[0060] Based on this, embodiments of this application provide a braking energy recovery system, method, and apparatus that can improve the user's driving experience.

[0061] For example, please refer to Figure 1This application provides a braking energy recovery system including an intelligent driving controller 100, a body controller 200, and a vehicle control unit (VCU) 300. The intelligent driving controller 100 acquires scene perception information of the vehicle. It also integrates a driving habit learning model to generate driving suggestions for future driving paths based on scene perception information and vehicle driving state information. The driving habit learning model is trained on the intelligent driving controller 100 or trained in the cloud and then deployed on the intelligent driving controller 100. The intelligent driving controller 100 can also be another domain controller with scene perception and model inference functions, such as an intelligent cockpit domain controller. The body controller 200 constructs a risk potential energy map corresponding to the vehicle's future driving path based on the vehicle's scene perception information. It also optimizes a multi-objective optimization function based on the risk potential energy map and driving suggestions. Optionally, the functions of the body controller 200 can be implemented through a VCU or a vehicle interface unit (VIU). The VCU 300 controls the vehicle to perform braking energy recovery using an optimal braking curve. Optionally, the VCU300's functions can also be implemented through a motor control unit (MCU). Optionally, the intelligent driving controller 100, the body controller 200, and the VCU300 can also accept instructions from the operator and flexibly configure the parameters involved in the braking energy recovery process based on the instructions, such as adjusting model parameters and weighting coefficients.

[0062] The braking energy recovery method provided in this application embodiment can be applied to the braking energy recovery system of the above-described embodiments, and can also be applied to other systems, devices, or equipment capable of realizing braking energy recovery. For example, please refer to... Figure 2 The braking energy recovery method provided in this application includes:

[0063] Step S201: The braking energy recovery system constructs a risk potential energy map corresponding to the vehicle's future driving path based on the vehicle's scene perception information.

[0064] The scene perception information includes the curvature of curves on the future driving path, the location and height of speed bumps, the status and countdown of traffic lights, the distance and relative speed between the vehicle and the vehicle in front, the level of road congestion, and the recommended speed for ramps.

[0065] For example, scene perception information can be obtained through cameras, radar, high-precision maps, vehicle-to-everything (V2X) technology, navigation, etc.

[0066] For example, an intelligent driving controller integrates visual recognition services (camera service), radar target information services (radar service), high-definition map services (HDMap service), V2X services, and navigation services. Visual recognition services (cameras) perceive lane lines, traffic signs (speed limits, traffic lights), and the types, outlines, and movement states of traffic participants (vehicles, pedestrians, bicycles). Radar target information services perceive the distance, relative speed, and azimuth between the vehicle and surrounding objects. High-definition map services provide information on lane lines (curvature, width), slope, heading, traffic sign positions, curbs, and medians. V2X services provide dynamic information from roadside units (RSUs) and other vehicles (such as blind spot incidents, traffic light status and countdowns, vulnerable road user warnings, road safety warnings, real-time traffic conditions, speed limit information, and high-definition map differential data). Navigation services provide future driving routes and estimated travel times.

[0067] Among them, vehicle V2X services refer to information interaction and application services provided by V2X for vehicles, road users and traffic management systems, specifically including communication and interaction between vehicles, between vehicles and people, between vehicles and infrastructure, and between vehicles and networks.

[0068] Traffic light status and countdown include the current color of the traffic light (red, green, yellow), the remaining time, and the next step (green light will turn red, or red light will turn green).

[0069] Road safety warnings include warning types such as accidents ahead, construction areas, road debris, fog, and icy water accumulation, as well as the corresponding geographical locations and impact ranges of each warning type.

[0070] Real-time traffic conditions include congestion level (average vehicle speed and congestion length on the road ahead, obtained from roadside radar / camera fusion perception) and queue information (number of vehicles waiting in line at the intersection ahead and queue length).

[0071] Speed ​​limit information includes dynamic speed limit signs, such as temporarily reduced speed limits during rainy or snowy weather or special periods.

[0072] High-precision map differential data includes changes in the geometry of roads along future driving routes, temporary detour information, etc., which are used to enhance or update high-precision maps for vehicles.

[0073] The risk potential energy map is used to characterize the correspondence between the driving speed of a location point on a future driving path and the risk potential energy of the driving scenario in which the location point is located.

[0074] For example, the driving scenario in which the target location is located is determined based on scene perception information, the target risk type associated with the driving scenario in which the target location is located is determined, the correspondence between the driving speed of the target location and the risk potential energy of the driving scenario in which the target location is located is constructed based on the target risk type, and a risk potential energy map is generated based on the correspondence between the driving speed of the target location and the risk potential energy of the driving scenario in which the target location is located.

[0075] Among them, the driving scenario is a single traffic scenario or a composite scenario of multiple traffic scenarios, and the target location point is each location point on the future driving path.

[0076] Optionally, traffic scenarios include curve scenarios, speed bump scenarios, traffic light scenarios, congestion scenarios, and ramp scenarios.

[0077] The target risk type is the risk type in the risk type dataset summarized from vehicle driving.

[0078] Optionally, the risk type dataset includes forward collision risk type, curve loss of control risk type, speed bump inconvenience risk type, traffic light violation risk type, and speeding violation risk type.

[0079] For example, when a camera detects a speed bump in a vehicle's path, the height or severity level of the speed bump is estimated using visual AI algorithms or high-precision map attributes to determine the type of discomfort risk. For instance, a monocular vision-based geometric feature estimation method can be used to extract the geometric features of the speed bump. A pre-trained deep learning model can then identify the speed bump region, and a pre-trained regression model can determine the severity level. Geometric features include visual height, aspect ratio, texture, and shadow analysis. The severity level is a continuous value or discrete level (e.g., mild, moderate, severe) related to the speed bump's height and bumpiness. Specifically, the image captured by the camera is pre-processed and feature-recognized (e.g., edge detection, texture analysis) to identify candidate regions that may include speed bumps. These candidate regions are input into a deep learning model for classification, outputting the speed bump region. The speed bump region and its geometric features are then input into a regression model, outputting the speed bump's severity level label (e.g., severity level type, score). For example, for speed bumps, firstly, based on images from a forward-looking camera, a deep learning model is used to identify the speed bump area and extract its geometric features (such as aspect ratio, texture, shadow, etc.). Then, a pre-trained regression model is used to map it to a severity level or estimate its height. Secondly, historical data is used to match the visual features of the current speed bump with historical response data such as suspension travel and vehicle vertical acceleration when passing through it, in order to calibrate and optimize the current estimation results. In addition, on road sections with high-precision map data, the pre-stored speed bump attribute information in the map can be directly called to improve the accuracy of severity level determination.

[0080] Risk potential is a parameter that measures the severity of the potential consequences when a target risk type "realizes" or "explodes." Examples include: the severity of personal safety and vehicle damage after a target risk type event occurs; the probability of the target risk type event occurring; and the length of the time window left for the vehicle or driver to react and avoid the target risk type.

[0081] In some embodiments, there are multiple target risk types. Based on the first mapping relationship and weight coefficients corresponding to each of the multiple target risk types, a correspondence is constructed between the driving speed of the target location and the risk potential energy of the driving scenario in which the target location is located.

[0082] The first mapping relationship is used to characterize the correspondence between the location point, the speed of the location point, and the risk potential energy of the location point under the corresponding target risk type.

[0083] For example, the risk potential energy corresponding to each target risk type is uniformly quantified into a normalized risk potential energy value between 0 and 1. The higher the risk, the closer the risk potential energy value is to 1.

[0084] For example, Table 1 shows the risk potential function for calculating the risk potential value for multiple target risk types.

[0085] Table 1: Risk Potential Function Table for Target Risk Types

[0086]

[0087] Where P_target represents the risk potential energy value, TTC represents the collision time when the vehicle collides with the target ahead at its current relative speed and direction (e.g., TTC is the relative distance / relative speed between the vehicle and the vehicle in front), and TTC_max is the preset maximum collision time threshold (e.g., a fixed value within 3-7 seconds, defining the vehicle's sensitivity to the risk of collision with the vehicle in front). v_current represents the vehicle's current speed, and v_safe_max represents the theoretical maximum safe speed at which the vehicle will not skid or lose control under specific curve curvature and road surface conditions. Specifically, v_safe_max is a parameter that varies with the target location point; different curves and different road conditions have different calculated values. The calculation formula is: v_safe_max = sqrt( μ represents the road adhesion coefficient (estimated via the electronic stability program (ESP)), and R represents the radius of curvature of the curve (provided by high-precision map services). v_comfort_max represents the maximum comfortable speed at which passengers will not experience significant discomfort when the vehicle traverses uneven road surfaces such as speed bumps. v_comfort_max is related to the characteristics of the speed bump (such as its height) identified by the scene perception service and is obtained by querying a preset "height-comfort speed" mapping table. f(time_to_red) represents a function that maps the remaining time before the traffic light turns red to a violation risk value; the closer to the red light's on time, the higher the risk potential value. v_legal_limit represents the maximum permissible speed for the future driving route, such as the speed limit, obtained through navigation map data, traffic sign recognition, V2X services, etc.

[0088] For example, the radius of curvature of a target location point x is obtained from a high-precision map, and v_safe_max(x) is calculated based on this radius. On a straight road, a large radius of curvature results in a high v_safe_max(x), and the potential energy value for the risk of losing control in a curve is almost zero. As the vehicle approaches the apex of the curve in its future travel path, the radius of curvature decreases, and v_safe_max(x) drops sharply, causing the potential energy value for the risk of losing control in a curve to peak at point x at the apex. With v_safe_max(x) (e.g., at the curve entrance) remaining constant, a higher v_current results in a higher potential energy value for the risk of losing control in a curve; if v_current is very low, the potential energy value for the risk of losing control in a curve approaches zero. Another example: the closer the vehicle's target location point x is to the stop line and the red light is about to turn on, the higher the potential energy value for traffic light violation at the target location point x. Yet another example: the higher the v_current and the lower the TTC (Traffic Traffic Convection) at the target location point x, the higher the potential energy value for a collision with the vehicle ahead.

[0089] Optionally, the weight coefficients corresponding to each of the multiple target risk types are default values ​​or preset values, which can be manually or automatically adjusted according to user needs.

[0090] In some embodiments, different target risk types correspond to different priorities, and different priorities have different weighting coefficients. As shown in Table 2:

[0091] Table 2: Weighting Coefficients for Target Risk Types

[0092]

[0093] For example, different target risk types and priorities corresponding to different driving scenarios are pre-configured.

[0094] In some embodiments, the process of determining the weighting coefficients corresponding to each of the multiple target risk types includes the following steps:

[0095] Step 11: Determine the target style adjustment factor based on the driver's driving style and the first mapping relationship dataset between driving style and style adjustment factor.

[0096] Optionally, driving styles include aggressive and conservative.

[0097] For example, driving style is determined based on vehicle driving parameters and a driving style prediction model. For instance, the driving habit learning model outputs aggressiveness: 0.8, comfort_preference: 0.9.

[0098] Different driving styles correspond to different driving style adjustment factors, as shown in Table 3:

[0099] Table 3: First Mapping Relationship Dataset

[0100]

[0101] Step 12: Determine the target environmental regulation factor based on the vehicle's environmental state and the second mapping relationship dataset between the environmental state and the environmental regulation factor.

[0102] The environmental condition includes the road surface adhesion condition and the visibility condition.

[0103] For example, the environmental state is obtained through the scene perception service module, such as road_friction: 0.6, which represents a low-friction road surface, and weather: rain, which represents a rainy day.

[0104] Different environmental conditions correspond to different environmental adjustment factors, as shown in Table 4:

[0105] Table 4: Second Mapping Relationship Dataset

[0106]

[0107] Step 13: Based on the target style adjustment factor, the target environment adjustment factor, and the basic weight coefficients corresponding to each of the multiple target risk types, determine the weight coefficients corresponding to each of the multiple target risk types.

[0108] For example, the product between the target style adjustment factor and the target environment adjustment factor is determined as the adjustment ratio. The adjustment ratio limit for each of the multiple target risk types is determined based on the risk level of each of the multiple target risk types. The weight coefficient corresponding to each of the multiple target risk types is determined based on the adjustment ratio, the adjustment ratio limit for each of the multiple target risk types, and the basic weight coefficient.

[0109] For example, as shown in Table 5:

[0110] Table 5:

[0111]

[0112] Step S202: The braking energy recovery system generates driving suggestions for future driving paths based on scene perception information and vehicle driving status information.

[0113] Optionally, the driving status information includes the driving speed and braking deceleration at each moment in the acquisition time sequence window corresponding to the current moment.

[0114] The acquisition timing window corresponding to the current moment is a time window consisting of the current moment and multiple moments preceding it. For example, the acquisition timing window corresponding to the current moment is a time window within 3 seconds preceding the current moment.

[0115] The driving recommendations are tailored to the driver's driving habits at the vehicle's speed.

[0116] Optionally, the driving recommendations include the target speed at the end point of the future driving path and the expected braking deceleration curve for the future driving path.

[0117] For example, the driving scenario of the target location is determined based on scene perception information, and driving suggestions for the future driving route are generated based on the driving scenario of the target location, driving status information and driving habit learning model.

[0118] Among them, the driving scenario is a single traffic scenario or a composite scenario of multiple traffic scenarios, the target location point is each location point on the future driving path, and the driving habit learning model is trained using historical sample data.

[0119] Optionally, the historical sample data includes: the driving speed and braking deceleration at each moment in the acquisition time window corresponding to the historical moment, the driving scenario at each location point on the future driving path corresponding to the historical moment, and the actual braking curve of the future driving path corresponding to the historical moment.

[0120] Among them, the braking curve on the future driving path is used to reflect the driving speed and braking deceleration at each position point on the future driving path.

[0121] Specifically, a driving habit learning model is trained based on historical sample data. The input of the driving habit learning model is driving scenario and driving state information, and the output of the driving habit learning model is driving suggestions for future driving routes.

[0122] For example, the training process of a driving habit learning model includes: inputting the driving speed and braking deceleration at each moment in the acquisition time sequence window corresponding to a fixed duration (e.g., 5 seconds) of historical moments into the driving habit learning model, the driving scenario at each position point on the future driving path corresponding to the historical moment, such as: [vehicle speed (t-5s), vehicle speed (t-4s), ..., vehicle speed (t0)], and the corresponding acceleration, yaw rate, and other data. For example, snapshots of predicted scenario information for the future driving path at each historical moment are released in chronological order. For instance, at t-5 seconds, the scenario list is: [Curve 300 meters ahead]; at t-3 seconds, the scenario list is updated to: [Curve 250 meters ahead, congestion starts 400 meters ahead]; at t-1 seconds, the scenario list is updated to: [Curve 200 meters ahead, congestion starts 350 meters ahead]; the scenario list is arranged chronologically as: [List (t-5s), List (t-3s), List (t-1s)], recording the dynamic evolution of the environmental cognitive information received by the driver before making a decision. The driving habit learning model, based on input data, predicts the driver's actual braking curve after the data acquisition window ends. For example, it predicts the vehicle's final stable speed within 2 seconds after the data acquisition window ends, and the actual deceleration curve generated by the driver's pedal operation after the data acquisition window ends. Mean squared error (MSE) is used as the loss function during training.

[0123] Optionally, the driving habit learning model employs a deep learning model that can effectively capture long-term temporal dependencies, such as the encoder part of a long short-term memory network (LSTM) or a self-attention model (transformer). Such models are adept at learning and inferring complex patterns from input sequence data.

[0124] In some embodiments, the training phase of the driving habit learning model is divided into a first phase and a second phase. In the first phase, training is performed based on historical sample data of a large-scale driver group covering various driving styles to obtain an initial prediction model capable of learning general driving behaviors. The initial prediction model is then deployed in real-world scenarios. The initial prediction model continuously collects the current driver's operation data and uses incremental learning or periodic fine-tuning techniques to optimize the parameters of the initial prediction model, enabling the general initial prediction model to quickly and continuously evolve into a driving habit learning model that accurately reflects the current driver's habits.

[0125] In some embodiments, to ensure the reliability of the driving habit learning model across all scenarios, optimized learning and training are performed for the target scenario. For example, for high-frequency scenarios, learning is accelerated by adjusting the dynamic learning rate, identifying scenario fingerprints, and maintaining a frequency table; for rare but important scenarios, learning strategies such as course learning, reinforcement learning simulation, and cross-user knowledge transfer are adopted.

[0126] Step S203: The braking energy recovery system optimizes the multi-objective optimization function based on the risk potential energy map and driving suggestions to obtain the optimal braking curve corresponding to the future driving path.

[0127] The multi-objective optimization function includes several optimization objectives: minimizing the total risk potential energy of the future driving path, maximizing the braking energy recovery efficiency of the future driving path, and minimizing the gap between the braking curve of the future driving path and the driving suggestion.

[0128] For example, the difference between the braking curve on the future driving path and the driving recommendation is determined based on a first deviation value and a second deviation value.

[0129] The first deviation value is the sum of the deviations between the driving speed and the target speed at each position point on the future driving path, and the second deviation value is the sum of the deviations between the braking deceleration and the expected braking deceleration at each position point on the future driving path. The expected braking deceleration at each position point on the future driving path is obtained by querying the expected braking deceleration curve.

[0130] In some embodiments, the process of constructing a multi-objective optimization function includes, but is not limited to, the following steps:

[0131] Step 21: Construct the first optimization term with the objective of minimizing the total risk potential energy of the future driving path.

[0132] Step 22: Construct a second optimization term with the objective of maximizing braking energy recovery efficiency.

[0133] The braking energy recovery efficiency is determined based on the braking deceleration at each point along the future driving path.

[0134] Specifically, if the average braking deceleration along the future driving path is too low, the energy recovered by the electric motor will be less; if the average braking deceleration along the future driving path is too high, friction braking may intervene, reducing braking regeneration efficiency. Therefore, an excessively low average braking deceleration falls within the range of efficient electric motor power generation (e.g., 0.1g). When the amount is 0.2g, the motor has higher power generation efficiency and higher braking energy recovery efficiency.

[0135] The smoother the braking deceleration curve along the future driving path (lower acceleration), the higher the braking energy recovery efficiency. This is because a smoother braking deceleration curve allows for more precise and smoother control of the motor torque, preventing the motor from hovering at inefficient points and avoiding frequent use of friction brakes for compensation due to control oscillations, resulting in higher braking energy recovery efficiency.

[0136] Step 23: Construct the third optimization term that minimizes the first deviation value.

[0137] Step 24: Construct the fourth optimization term that minimizes the second deviation value.

[0138] Step 25: Construct a multi-objective optimization function based on the first optimization term, the second optimization term, the third optimization term, and the fourth optimization term, as well as their respective optimization weights and constraints.

[0139] In some embodiments, the driving suggestion includes the desired deceleration curve shape, constructing a fifth optimization term with the objective of minimizing the rate of change of braking deceleration at each position point on the future driving path, and constructing a multi-objective optimization function based on the first, second, third, fourth, and fifth optimization terms, as well as the optimization weights and constraints of the first, second, third, fourth, and fifth optimization terms.

[0140] The optimization weight of the fifth optimization term is taken from the weight change range corresponding to the shape of the desired deceleration curve.

[0141] For example, a cost function is constructed based on the first, second, third, and fourth optimization terms. Exemplarily, the cost function is expressed as the following formula:

[0142] .

[0143] in, It represents the total risk potential energy, and characterizes the correspondence between the driving speed v of a position point on the future driving path and the risk potential energy of the driving scenario in which position point x is located. It represents the acceleration at each point on the future driving path, reflecting the rate of change of deceleration, and is derived from the driving speed curve of the future driving path. This represents the target speed output by the driving habit learning model. This indicates the speed at each point along the future driving path. This represents the first deviation value. This represents the expected braking deceleration output by the driving habit learning model. This represents the braking deceleration at each point along the future driving path. This represents the second deviation value. This represents the optimization weights used to optimize jerk. This represents the optimization weight corresponding to the first deviation value. This represents the optimization weight corresponding to the second deviation value.

[0144] By employing numerical optimization algorithms (such as model predictive control, MPC) to solve the cost function in real time, a smooth braking deceleration curve with minimal cost can be output. This braking deceleration curve has a smaller total risk potential energy, higher braking energy recovery efficiency, and can better match the driver's driving habits, resulting in a better user experience.

[0145] In some embodiments, the desired deceleration curve shape is determined, and different deceleration curve shapes correspond to different ranges of acceleration values ​​and different optimization weight values. For example, when the deceleration curve shape is linear, the optimization weight... When the weight is 0.5 and the deceleration curve is S-shaped, the optimization weight is... The optimal weight is 1.2 when the deceleration curve is S-shaped. Larger. For example, the jerk when the deceleration curve is linear is higher than the jerk when the deceleration curve is S-shaped. For example, when the deceleration curve is in its initial stage, a stronger deceleration constraint (higher deceleration) is set in the early stages of optimization.

[0146] In some embodiments, safety constraints are introduced during the optimization process of constructing a multi-objective optimization function. For example, the vehicle's speed at each location point on the future travel path is less than or equal to the maximum safe speed corresponding to that location. The maximum safe speed is calculated in real time based on factors such as the curvature of the curves and the road surface adhesion coefficient of the future travel path. Another example is that the vehicle's speed at each location point on the future travel path is less than or equal to the speed limit corresponding to that location. The speed limit is obtained based on navigation or traffic sign recognition.

[0147] For example, in a scenario involving sudden congestion on a high-speed curve, the risk of a collision with the vehicle ahead and the risk of losing control in the curve are calculated. The risk of a collision with the vehicle ahead has a higher priority than the risk of losing control in the curve. The total risk potential is dominated by the risk of a collision with the vehicle ahead when the vehicle is close to the area in front. Under the constraints of "vehicle speed entering the curve ≤ safe speed in the curve" and "vehicle speed ≤ road speed limit", the optimizer solves for the trajectory that minimizes the cost function J, and plans a strong deceleration curve that first meets the requirements of following and avoiding collisions, while ensuring that the vehicle speed has naturally decreased to below the safe speed for cornering at the curve entrance. Finally, it outputs a smooth "S-shaped" curve that is sharp at first and then slows down, achieving safe, comfortable and efficient braking control.

[0148] Specifically, the scenario of encountering congestion on a high-speed curve includes Scenario 1 and Scenario 2. For Scenario 1 (static), based on the curve radius R = 200 meters and a dry road surface (μ = 0.8), the calculated physical safe speed v_safe_max ≈ 80 km / h. For Scenario 2 (dynamic), there is congestion - a slow-moving flow of traffic 80 meters ahead, with an average speed of 30 km / h, and the current relative speed TTC (Time to Collision) is rapidly decreasing. First, when the system simultaneously perceives Scenario 1 and Scenario 2, it calculates the total risk potential energy for both scenarios. The risk potential energy for Scenario 1 is calculated using (v / 80)², requiring the vehicle's speed to not exceed 80 km / h to prevent loss of control. The risk potential energy for Scenario 2 is calculated using 1 - (TTC / TTC_max), requiring the vehicle to decelerate to avoid a rear-end collision. Under the constraint of a speed ≤ 80 km / h, the MPC optimizer solves the cost function J and outputs an optimal future speed-distance trajectory curve. For example, Table 6:

[0149] Table 6: Future Vehicle Speed-Distance Trajectory Curve Data Table

[0150]

[0151] The curves shown in Table 6 indicate that in the early stage of braking (x=80-60m), the potential energy of collision risk dominates, requiring gentle deceleration to close the following distance. In the middle stage of braking (x=60-20m), the potential energy of cornering risk increases sharply, superimposing with the potential energy of collision risk, resulting in the strongest deceleration to simultaneously satisfy "safe entry into the corner" and "avoidance of rear-end collisions." In the late stage of braking (x=20-0m), the potential energy of risk decreases, requiring gentle deceleration to smoothly pass through the apex of the corner.

[0152] Step S204: The brake energy recovery system uses the optimal braking curve to control the vehicle to recover brake energy.

[0153] For example, the regenerative braking system sends control commands to the motor controller, hydraulic brake controller, etc., such as converting torque requests into motor commands (motorcontrolservice) and coordinating hydraulic brake commands (brakecontrolservice). When the motor's regenerative torque is insufficient to achieve the target deceleration, it requests the ESP to provide braking force compensation, while simultaneously reducing the motor torque request accordingly to ensure accurate and smooth total braking force.

[0154] For example, regarding the future speed-distance trajectory curve data shown in Table 6, at trajectory point x=40m, the target deceleration a=-1.5m / s², and the vehicle mass m=1500kg, first calculate the required total braking force F_total=m |a|=1500 1.5 = 2250N. Then, the maximum regenerative braking capacity of the vehicle's motor is checked. Based on the vehicle's current speed (70km / h) and battery SOC (75%), a table is consulted to determine that the motor can provide a maximum braking force of F_motor_max = 1200N. The mechanical braking compensation is calculated as: F_brak = F_total - F_motor_max = 2250 - 1200 = 1050N. Next, the motor torque command Motor_Torque = -(F_motor_max) is generated. Wheel_Radius) = -(1200N 0.3m) = -360Nm (the negative sign indicates negative torque, i.e., energy recovery), and the ESP braking command: converting F_brake = 1050N into the corresponding hydraulic braking pressure (e.g., 25Bar). Finally, the motor torque and mechanical braking are coordinated. For example, if the target torque is -360Nm, the actual control command sent will smoothly transition with a certain gradient (e.g., 500Nm / s) to avoid shock. Another example: by precisely controlling the hydraulic pressure through ESP, closely connecting it with the motor torque, the driver is ensured to feel a single, smooth total braking force. The total braking force also includes the braking force provided by air resistance and ground friction. If the target deceleration is less than the deceleration provided by air resistance and ground friction, the motor does not need to provide braking force.

[0155] like Figure 3 As shown in the figure, the braking energy recovery architecture provided in this application includes a scene perception and service module 10, an energy management decision and service module 20, a vehicle control execution service module 30, a driving style cloning and service module 40, and a service registration and discovery center 50. The braking energy recovery architecture is based on a service-oriented architecture. The regenerative braking energy recovery system is implemented using a Service-Oriented Architecture (SOA). SOA modularizes different functional services of an application. These services communicate and combine over a network through standardized interfaces and protocols to build complex applications. Each service can be developed, deployed, and maintained independently, thereby improving the flexibility, scalability, and reusability of the regenerative braking energy recovery system.

[0156] For example, such as Figure 4As shown, the scene perception and service module 10 continuously perceives the scene and continuously publishes a scene list to the energy management decision and service module 20. Upon receiving the scene list, the energy management decision and service module 20 triggers a decision, invoking driving suggestions provided by the driving style cloning and service module 40. The driving style cloning and service module 40 returns the target speed, desired braking deceleration, etc. After receiving the driving suggestions, the energy management decision and service module 20 combines multi-scene fusion and safety verification to make a final decision and plan, generating the optimal braking curve corresponding to the future driving path, and then generating the control command corresponding to the optimal braking curve. The vehicle control execution service module 30 executes the control command and provides real-time feedback on the actual vehicle status. The energy management decision and service module 20 then adjusts the control command based on the feedback status to achieve continuous control.

[0157] Specifically, the Scene Perception and Service Module 10, Energy Management Decision and Service Module 20, Vehicle Control Execution Service Module 30, and Driving Style Cloning and Service Module 40 are loosely coupled through the Service Registration and Discovery Center 50. The Scene Perception and Service Module 10 runs on the domain controller (such as the Intelligent Driving Domain Controller) as a collection of multiple atomic services, including providing visual recognition results, providing radar target information, providing high-precision map static information, receiving roadside information, and providing navigation guidance information. The Scene Perception and Service Module 10 performs data fusion internally to generate a unified service interface, PredictiveScenarioService, which can publish multiple scene information simultaneously (such as simultaneously publishing curve, congestion, and speed bump information).

[0158] The Driving Style Cloning and Service Module 40 runs on the intelligent cockpit domain controller or cloud for training and vehicle-side inference. It trains a deep learning model (such as an LSTM network) by learning real-world driver operation data (e.g., pedal signals, vehicle speed changes) in various scenarios over a long period, learning the driver's actual behavioral habits. The Energy Management Decision and Service Module 20 provides a Driving Style Model Service and a multi-scenario collaborative query interface (QueryPreferredDeceleration). For example, it queries the desired braking deceleration, taking vehicle state data (CurrentVehicleState) and a scene list (SceneList) as input, and returning a comprehensive deceleration scheme (ComprehensiveDecelerationProfile). It also provides a Driving Style Tag Query Interface (GetDrivingStyleProfile), returning a driver style tag (DrivingStyleProfile). CurrentVehicleState is a structure containing dynamic information about the vehicle at the time of query, such as current speed, current acceleration, yaw rate, current vehicle mass, remaining battery percentage, and current gear. SceneList is an array of SceneObjects. Each SceneObject contains the scene type, distance to the scene, estimated arrival time, and scene-specific attributes. Scene-specific attributes vary depending on the scene type. For example, a curve scene has attributes like road curvature, a speed bump scene has the estimated speed bump height, a traffic light scene has the remaining time of the traffic light, and a congested scene has the average speed of the congested traffic ahead. ComprehensiveDecelerationProfile is a structure containing parameters characterizing the driver's style, used by the decision-making module for refined trajectory planning. These include: target speed (the speed the driver ultimately wants to achieve in the given scene, such as coasting to 48 km / h on a 40 km / h speed limit ramp) and preferred deceleration (the driver's preferred deceleration value in the given scene, such as -0.8 m / s²). 2The deceleration curve hint (an enumerated value or parameter indicating the preferred deceleration curve shape, such as linear deceleration, gentle S-shaped deceleration, or initial acceleration) is displayed. The DrivingStyleProfile is a structure containing the driver's driving style (e.g., aggressiveness, floating-point number, range [0,1]), comfort preference (floating-point number, range [0,1]), efficiency focus (floating-point number, range [0,1]), and last update timestamp (last_updated). The last update timestamp is used to handle complex scenarios. The interface of the Energy Management Decision and Service Module 20 receives the current vehicle state and a list of scenarios, and returns parameters such as target vehicle speed and desired braking deceleration.

[0159] The Energy Management Decision and Service Module 20 runs on the vehicle domain controller, VCU, or VIU, and subscribes to the PredictiveScenarioService published by the Scene Awareness and Service Module 10 to obtain driving scenarios in real time. It also subscribes to the DrivingStyleModelService published by the Driving Style Cloning and Service Module 40 to obtain target speeds and decelerations that match the driver's habits.

[0160] The vehicle control execution service module 30 runs on the VCU or MCU, subscribes to the braking energy recovery strategy published by the energy management decision and service module 20, and generates control commands to perform braking energy recovery.

[0161] like Figure 5 As shown in the embodiments of this application, another method for regenerative braking includes:

[0162] Step S501: When the braking energy recovery system is started, each service registers with the registration center and subscribes to the required services, and establishes a communication link.

[0163] Step S502: The scene perception and service module continuously scans the environment in front of the vehicle to identify driving scenarios such as curves, speed bumps, red lights, traffic jams, and ramps.

[0164] Step S503: After identifying the driving scenario, the energy management decision and service module executes a unified optimization process, queries the driving style cloning and service module for driving suggestions, and then performs optimization calculations in conjunction with the risk potential energy map to generate the optimal braking curve corresponding to the future driving path.

[0165] Step S504: The vehicle control execution service module generates torque control commands based on the optimal braking curve and sends the torque control commands to the motor and braking system for brake energy recovery.

[0166] Step S505: Continuous monitoring. If the deceleration target is not reached or the scene does not disappear, return to step S502 to continue sensing and adjustment, forming a closed loop.

[0167] Figure 6 This diagram illustrates a driving speed curve provided in an embodiment of this application. The horizontal axis represents the vehicle's position x on its future driving path, in meters (m); the vertical axis represents the driving speed v at that position on the future driving path, in kilometers per hour (km / h). After sensing a curve, the vehicle determines driving suggestions through a driving habit learning model. Then, based on the driving suggestions and a risk potential energy map, it determines the optimal braking curve and initiates energy recovery based on this optimal braking curve. Initiating smooth deceleration and energy recovery in advance allows the vehicle's speed to gradually decrease to a safe value before reaching the curve. The curve starts from the current driving speed, and as the vehicle approaches the curve, the speed smoothly decreases, ultimately reaching a safe cornering speed at the entry point.

[0168] like Figure 7 As shown in the embodiment of this application, a deceleration curve optimization method includes:

[0169] Step S701: Calculate the risk potential energy for each scenario.

[0170] Step S702: Identify the scenario with the greatest risk potential as the primary risk scenario, and the remaining scenarios as secondary risk scenarios.

[0171] Step S703: Solve the cost function with the main risk scenario as the optimization objective and the secondary risk scenario as the constraint condition, and output the deceleration curve. For example, when the curve scenario is the secondary risk scenario, the safe speed for curves is used as the constraint condition, and the vehicle's speed when entering the curve is less than or equal to the safe speed for curves.

[0172] Figure 8 This is a schematic diagram of another driving speed curve provided in an embodiment of this application. The horizontal axis represents the vehicle's position x on the future driving path, in meters (m); the vertical axis represents the driving speed v at that position on the future driving path, in kilometers per hour (km / h). Figure 8As shown, the vehicle encounters two different curves when passing over a high speed bump and a low speed bump, with the high speed bump being higher than the low speed bump. The high speed bump results in a greater deceleration, causing the vehicle to slow down earlier and more rapidly, ensuring a lower speed when passing over it. The low speed bump, on the other hand, has a gentler deceleration, prioritizing comfort. Braking recovery is earlier and stronger over the high speed bump, while it is later and gentler over the low speed bump. The difference in the starting points of the two curves reflects the automatic quantification and response to the severity (height) of the scenario based on risk potential energy. After sensing a speed bump, the vehicle determines driving suggestions through a driving habit learning model, then determines the optimal braking curve based on the driving suggestions and the risk potential energy map. Energy recovery begins based on this optimal braking curve, resulting in a lower vehicle speed when passing over the speed bump, improving driving comfort and safety. Specifically, the scenario perception and service module identifies a high speed bump (8cm high) 80 meters ahead of the vehicle and publishes information about the high speed bump (including distance and height). The energy management decision and service module sends a list of scenarios containing information about high speed bumps to the driving style cloning and service module, which then returns the driver's preferred deceleration (e.g., -0.9 m / s²). 2 (Driving style is mild). The energy management decision and service module calculates the risk potential energy value based on the high speed bump information, constructs a cost function based on the risk potential energy value and the corresponding optimization weight (comfort weight), and the optimizer solves the cost function to obtain a deceleration (e.g., -1.2m / s²) that balances mild driving style and comfort for energy recovery, ensuring that the vehicle passes over the high speed bump at a low and comfortable speed. During energy recovery, the vehicle control execution service module receives the control commands generated by the energy management decision and service module and controls the motor to provide negative torque.

[0173] Figure 9 This is a schematic diagram of another driving speed curve provided in an embodiment of this application. The horizontal axis represents the position x of the vehicle on the future driving path, in meters (m); the vertical axis represents the driving speed v of the position point on the future driving path, in kilometers per hour (km / h). Figure 9The speed curve in the diagram corresponds to a highway exit ramp scenario. After the vehicle receives an exit ramp prompt from the navigation system, a driving habit learning model determines driving suggestions. Then, based on these suggestions and a risk potential energy map, an optimal braking curve is determined. Energy recovery begins based on this optimal braking curve, causing the vehicle to gradually decrease from its highway cruising speed (120 km / h) to the ramp speed limit (60 km / h) before entering the ramp's starting point. Specifically, while cruising at high speed (120 km / h), the navigation system provides an early exit ramp prompt. The driving style cloning and service module obtains the driver's habit of passing through ramps at speeds higher than the speed limit (60 km / h), such as 65 km / h. However, the ramp speed limit is 60 km / h. Therefore, a smooth deceleration curve is planned with the speed limit as a constraint, and energy recovery begins early, ensuring the vehicle speed smoothly decreases to 60 km / h upon reaching the ramp's starting point. For example, the scene perception and service module issues an exit ramp prompt when the vehicle is 1.5 kilometers (adjustable) away from the ramp; as it approaches the ramp, it senses the ramp curvature and the 40km / h speed limit sign, merges the ramp curvature and the 40km / h speed limit sign into composite scene information, and issues it. The energy management decision and service module sends the scenario "navigation exit ramp, curvature X, speed limit 40" as a scene list to the driving style cloning and service module. The driving style cloning and service module returns the driver's habitual target speed as 48km / h and a preference for two-stage smooth deceleration. The energy management decision and service module calculates the total risk potential energy based on parameters such as ramp curvature. Since the weight coefficients of stability safety (P1) and legal regulations (P2) are higher than those of traffic efficiency (P4), the calculated total risk potential energy value is higher in the ramp area. When the optimizer solves the driving speed curve with speed limit information as a constraint, it is subject to the combined constraints of high-weight risk potential energy and "vehicle speed ≤ 40km / h", generating a driving speed curve that smoothly decelerates to 40km / h. At the same time, the driving speed curve conforms to the driver's habit of "two-stage smooth deceleration" to the greatest extent.

[0174] like Figure 10 As shown, the braking energy recovery device provided in this application may include a processing module 1001 and a braking module 1002. The processing module 1001 is used to perform... Figure 2 In the illustrated method, steps S201, S202, and S203 are performed by the braking module 1002. Figure 2 The illustrated method includes step S204.

[0175] The foregoing mainly describes the solutions provided by the embodiments of this application from the perspectives of methods, systems, and apparatus. To achieve the above functions, the regenerative braking device includes corresponding hardware structures and / or software modules for performing each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0176] This application embodiment can, based on the above-described regenerative braking method, exemplarily divide the regenerative braking device into functional modules. For example, a vehicle speed limiting control system or vehicle speed limiting device may include functional modules corresponding to each functional division, or two or more functions may be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. It should be noted that the module division in this application embodiment is illustrative and only represents one logical functional division; in actual implementation, there may be other division methods.

[0177] This application also provides a computer-readable storage medium storing at least one computer program, which is loaded and executed by a processor to implement the braking energy recovery method provided in the above-described method embodiments.

[0178] Optionally, the computer-readable storage medium may be a non-transitory computer-readable storage medium, such as a read-only memory (ROM), random access memory (RAM), magnetic tape, floppy disk, and optical data storage device.

[0179] This application also provides a computer program product, which includes a computer program or instructions. When the computer program or instructions are executed by a processor, the braking energy recovery method provided in the above-described method embodiments is implemented.

[0180] It should be noted that when one or more instructions in the computer-readable storage medium or computer program product are executed by the processor of a computing device, they implement the various processes of the above-described method embodiments and achieve the same technical effects as the above-described methods. To avoid repetition, they will not be described again here.

[0181] Through the above description of the embodiments, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0182] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another apparatus, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or units may be electrical, mechanical, or other forms.

[0183] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0184] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0185] It should be understood that the application of this application is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims. Those skilled in the art can understand that implementing all or part of the processes of the above embodiments and making equivalent changes according to the claims of this application still fall within the scope of this application.

Claims

1. A method for recovering braking energy, characterized in that, The method includes: Based on the vehicle's scene perception information, a risk potential energy map corresponding to the vehicle's future driving path is constructed; wherein, the risk potential energy map is used to characterize the correspondence between the driving speed of a position point on the future driving path and the risk potential energy of the driving scene in which the position point is located. Based on the scene perception information and the vehicle's driving status information, a driving suggestion for the future driving route is generated; wherein, the driving suggestion conforms to the driver's driving habits at driving speed. Based on the risk potential energy map and the driving suggestion, the multi-objective optimization function is optimized and solved to obtain the optimal braking curve corresponding to the future driving path; wherein, the multiple optimization objectives in the multi-objective optimization function include: minimizing the total risk potential energy of the future driving path, maximizing the braking energy recovery efficiency of the future driving path, and minimizing the difference between the braking curve of the future driving path and the driving suggestion; The vehicle is controlled to perform brake energy recovery using the optimal braking curve.

2. The braking energy recovery method according to claim 1, characterized in that, The construction of a risk potential energy map corresponding to the vehicle's future driving path based on the vehicle's scene perception information includes: Based on the scene perception information, the driving scene in which the target location point is located is determined; wherein, the driving scene is a single traffic scene or a composite scene of multiple traffic scenes; the target location point is each location point on the future driving path; Determine the target risk type associated with the driving scenario in which the target location is located; wherein, the target risk type is the risk type in the risk type dataset summarized by vehicle driving; Based on the target risk type, a correspondence is constructed between the driving speed at the target location and the risk potential energy of the driving scenario in which the target location is located; The risk potential energy map is generated based on the correspondence between the driving speed at the target location and the risk potential energy of the driving scenario in which the target location is located.

3. The braking energy recovery method according to claim 2, characterized in that, The traffic scenarios include: curve scenarios, speed bump scenarios, traffic light scenarios, congestion scenarios, and ramp scenarios; the risk type dataset includes: forward collision risk type, curve loss of control risk type, speed bump inconvenience risk type, traffic light violation risk type, and speed limit violation risk type.

4. The braking energy recovery method according to claim 2 or 3, characterized in that, The number of target risk types is multiple; the step of constructing a correspondence between the driving speed of the target location and the risk potential energy of the driving scenario in which the target location is located, based on the target risk types, includes: Based on the first mapping relationship and weight coefficients corresponding to multiple target risk types, a correspondence is constructed between the driving speed of the target location and the risk potential energy of the driving scenario in which the target location is located; The first mapping relationship is used to characterize the correspondence between a location point, the speed at which the location point travels, and the risk potential energy of the location point under the corresponding target risk type.

5. The braking energy recovery method according to claim 4, characterized in that, The process of determining the weight coefficients corresponding to each of the multiple target risk types includes: Based on the driver's driving style and the first mapping relationship dataset between driving style and style adjustment factor, a target style adjustment factor is determined; wherein, the driving style includes aggressive and conservative. Based on the vehicle's environmental state and a second mapping relationship dataset between the environmental state and environmental adjustment factors, the target environmental adjustment factor is determined; wherein, the environmental state includes road surface adhesion state and visibility state; Based on the target style adjustment factor, the target environment adjustment factor, and the basic weight coefficients corresponding to the multiple target risk types, the weight coefficients corresponding to the multiple target risk types are determined.

6. The braking energy recovery method according to claim 5, characterized in that, The determination of the weight coefficients corresponding to each of the multiple target risk types based on the target style adjustment factor, the target environment adjustment factor, and the basic weight coefficients corresponding to each of the multiple target risk types includes: The product of the target style adjustment factor and the target environment adjustment factor is determined as the adjustment ratio; Based on the risk level of each of the multiple target risk types, determine the adjustment ratio limit for each of the multiple target risk types; Based on the adjustment ratio, the adjustment ratio limits of each of the multiple target risk types, and the basic weight coefficient, the weight coefficient corresponding to each of the multiple target risk types is determined.

7. The braking energy recovery method according to claim 1, characterized in that, The driving recommendations include: the target speed at the end point of the future driving path and the expected braking deceleration curve of the future driving path.

8. The braking energy recovery method according to claim 1 or 7, characterized in that, The driving status information includes the driving speed and braking deceleration at each moment in the acquisition time sequence window corresponding to the current moment; the acquisition time sequence window corresponding to the current moment is a time window composed of the current moment and multiple moments before it; The step of generating driving suggestions for the future driving route based on the scene perception information and the driving status information includes: Based on the scene perception information, the driving scene in which the target location point is located is determined; wherein, the driving scene is a single traffic scene or a composite scene of multiple traffic scenes; the target location point is each location point on the future driving path; Based on the driving scenario where the target location is located, the driving status information, and the driving habit learning model, a driving suggestion for the future driving route is generated; The driving habit learning model is trained using historical sample data. The historical sample data includes: the driving speed and braking deceleration at each moment in the acquisition time window corresponding to the historical moment, the driving scenario at each position point on the future driving path corresponding to the historical moment, and the actual braking curve of the future driving path corresponding to the historical moment.

9. The braking energy recovery method according to claim 7, characterized in that, The braking curve on the future driving path is used to reflect the driving speed and braking deceleration at each position point on the future driving path; The difference between the braking curve on the future driving path and the driving suggestion is determined based on a first deviation value and a second deviation value. The first deviation value is the sum of the deviations between the driving speed at each location point on the future driving path and the target speed; The second deviation value is the sum of the deviations between the braking deceleration at each position point on the future driving path and the desired braking deceleration; The desired braking deceleration at each location point on the future driving path is obtained by querying the desired braking deceleration curve. The process of constructing the multi-objective optimization function includes: Construct a first optimization term with the objective of minimizing the total risk potential energy of the future driving path; A second optimization term is constructed with the objective of maximizing braking energy recovery efficiency; wherein the braking energy recovery efficiency is determined based on the braking deceleration at each position point on the future driving path; Construct a third optimization term that minimizes the first deviation value; Construct a fourth optimization term that minimizes the second deviation value; The multi-objective optimization function is constructed based on the first optimization term, the second optimization term, the third optimization term, and the fourth optimization term, as well as the optimization weights and constraints of the first optimization term, the second optimization term, the third optimization term, and the fourth optimization term.

10. The braking energy recovery method according to claim 9, characterized in that, The driving recommendations include the desired deceleration curve shape; The construction of the multi-objective optimization function based on the first optimization term, the second optimization term, the third optimization term, and the fourth optimization term, as well as the optimization weights and constraints of each of the first optimization term, the second optimization term, the third optimization term, and the fourth optimization term, includes: A fifth optimization term is constructed with the objective of minimizing the rate of change of braking deceleration at each location point on the future driving path; Based on the first optimization term, the second optimization term, the third optimization term, the fourth optimization term, and the fifth optimization term, as well as the optimization weights and constraints of the first optimization term, the second optimization term, the third optimization term, the fourth optimization term, and the fifth optimization term, the multi-objective optimization function is constructed. The optimization weight of the fifth optimization term is taken from the weight change range corresponding to the desired deceleration curve shape.

11. A braking energy recovery device, characterized in that, include: The processing module is used to construct a risk potential energy map corresponding to the future driving path of the vehicle based on the vehicle's scene perception information; wherein, the risk potential energy map is used to characterize the correspondence between the driving speed of the position point on the future driving path and the risk potential energy of the driving scene in which the position point is located. The processing module is further configured to generate a driving suggestion for the future driving path based on the scene perception information and the vehicle's driving status information; wherein the driving suggestion conforms to the driver's driving habits at driving speed. The processing module is further configured to optimize and solve a multi-objective optimization function based on the risk potential energy map and the driving suggestion to obtain the optimal braking curve corresponding to the future driving path; wherein, the multiple optimization objectives in the multi-objective optimization function include: minimizing the total risk potential energy of the future driving path, maximizing the braking energy recovery efficiency of the future driving path, and minimizing the difference between the braking curve of the future driving path and the driving suggestion; A braking module is used to control the vehicle to perform brake energy recovery using the optimal braking curve.

12. A vehicle, characterized in that, The vehicle includes the brake energy recovery device as described in claim 11.

Images