A PHEV energy management method and system based on visual traffic information
The traffic information is obtained through the on-board camera and combined with the YOLO image recognition model and optimization control strategy, the problems of PHEV in accurate positioning and real-time power flow optimization are solved, and energy management efficiency and battery life are improved.
Patent Information
- Application Number
- CN202210842569.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-07-18
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-07-18
AI Technical Summary
When obtaining and utilizing traffic information, existing hybrid vehicle PHEVs are difficult to achieve accurate lane-level positioning and real-time dynamic power flow optimization, resulting in low energy management efficiency.
The vehicle-mounted front-view camera is used to obtain traffic information, and the effective information is extracted through the pre-trained YOLO image recognition model. Combined with QP secondary planning and PMP minimum value principle, an optimal control model is built, and iterative optimization is used to achieve engine requirements. Optimal control of torque.
Improves the accuracy and real-time performance of PHEV energy management, optimizes the longitudinal acceleration and fuel consumption of the vehicle, and extends battery life.
Smart Images

Figure CN115257692B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of hybrid electric vehicle (PHEV) energy management, and in particular to a PHEV energy management method and system based on visual traffic information. Background Art
[0002] As an important part of energy-saving and new energy vehicles, hybrid electric vehicles (PHEVs) usually have a power system structure with more than two power sources. The role of the hybrid electric vehicle energy management system is to optimize the distribution of power flow through the coordinated control of multiple power sources, thereby improving and enhancing the working performance of the entire vehicle.
[0003] In recent years, with the rise and research boom of technologies such as smart cars and vehicle networking, technologies such as vehicle-road collaboration and intelligent traffic information have been vigorously developed and applied, which can enable hybrid vehicles to obtain traffic information data such as road information and working condition information to reach the target location, laying the foundation for improving the working efficiency of the energy management system. Hybrid vehicles have the characteristics of a drive system with multiple power sources. According to the integrated traffic information such as road and vehicle status, they can actively and effectively distribute power between different power sources in real time to meet the multi-objective control needs of the vehicle under different working conditions. Realizing multi-objective control and optimization of hybrid vehicle energy management that integrates traffic information has practical and important significance for reducing vehicle energy consumption and improving drivability.
[0004] The integration of traffic information can increase the richness of driving condition data and help optimize the energy management efficiency of hybrid vehicles. However, the extraction and integration of traffic flow information has the problem of large data volume and high dimension. In current applications, most car companies load traffic information into the on-board ECU through on-board navigation, and based on regularized control strategies, change the energy allocation ratio in the control strategy to achieve dynamic adjustment of power flow.
[0005] However, the above technology has the following problems:
[0006] 1) Based on the accuracy of GPS positioning of ordinary vehicles, it is difficult to achieve accurate lane-level positioning and it is difficult to accurately obtain traffic flow information before and after the current lane;
[0007] 2) The traffic information obtained by the in-vehicle navigation is relatively limited, and most of it is the traffic information of the current road section. It is difficult to make full use of traffic information such as traffic light phase lights to achieve real-time dynamic optimization of PHEV power flow distribution. Summary of the invention
[0008] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide a PHEV energy management method and system based on visual traffic information with accurate traffic information acquisition and efficient PHEV power flow distribution.
[0009] The purpose of the present invention can be achieved by the following technical solutions:
[0010] According to a first aspect of the present invention, a PHEV energy management method based on visual traffic information is provided, the method comprising the following steps:
[0011] Step S1, obtaining image information in front of the target vehicle in real time through the on-board front-view camera, and using the YOLO image recognition model with pre-trained visual traffic information weights to extract effective traffic information from the image information;
[0012] Step S2: The extracted effective traffic information is used as the real-time state quantity and constraint quantity input, trajectory planning is performed within a given prediction time domain based on the QP quadratic programming strategy, and the target vehicle speed trajectory within the current prediction time domain is output;
[0013] Step S3, based on the target vehicle speed trajectory in the current prediction time domain obtained by planning, the required power at the current moment is calculated, and the battery current input is used as one of the cost indicators to construct an optimal control model based on the PMP minimum principle, and the optimal control amount of the engine required torque at the current moment is solved by Hamiltonian function optimization;
[0014] Step S4: adopt the model predictive control MPC rolling optimization architecture to solve the optimal control quantity sequence of the engine demand torque at each forward step of the iteration, and output the first control quantity in the solved control quantity sequence to the vehicle control unit.
[0015] Preferably, the effective traffic information in step S1 includes the longitudinal distance to the vehicle ahead, the state and distance of the traffic light at the intersection ahead, and the distance and speed limit value information of the speed limit sign ahead.
[0016] Preferably, the pre-training of the YOLO image recognition model for visual traffic information weights in step S1 has a pre-training process for visual traffic information weights as follows:
[0017] 1) Modeling the conventional traffic environment and the target vehicle in a three-dimensional simulation environment; wherein a forward-looking camera is installed directly in front of the front windshield of the target vehicle to capture the traffic scene ahead;
[0018] 2) Set a vehicle-following scenario, with the speed trajectory of the front vehicle fixed and the rear vehicle being the target vehicle with a camera installed. Set the simulation step length and perform a vehicle-following simulation test under the set working conditions. During the simulation process, save the front view image taken by the front view camera in real time according to the simulation step length.
[0019] 3) Pre-calibrate the forward-looking simulation image and classify it based on the distance to the vehicle ahead, the distance to the traffic light ahead, and the distance to the speed limit sign ahead;
[0020] 4) Importing the classified images into the YOLO v5 image recognition model according to categories, and fitting neural network weight files of different categories through YOLO v5 image recognition training and offline iterative training; the neural network is divided into a first neural network, a second neural network and a third neural network according to categories; wherein the first neural network outputs the front vehicle distance L_fr, the second neural network outputs the traffic light distance L_lgt and the traffic light state Sta_lgt, and the third neural network outputs the front speed limit sign distance L_lim and the speed limit value V_lim.
[0021] Preferably, the mathematical expression of the QP quadratic programming model in step S2 is:
[0022] 1) Quadratic programming objective function:
[0023]
[0024] In the formula, x is the optimization parameter, which refers to the acceleration trajectory to be planned; H and Q are coefficient matrices, and their expressions are:
[0025]
[0026]
[0027] Where M is the mass of the vehicle, A is the windward cross-sectional area of the vehicle, and C D is the drag coefficient, f is the road friction coefficient, and ρ is the air density; is an N-order diagonal matrix, N is the prediction time domain; ψ=ones(N), is an N-order vector;
[0028] 2) Constraints:
[0029]
[0030] ψ·v min ≤ψ·v0+ψ·a≤ψ·v max
[0031] ψ·a min ≤a≤ψ·a max
[0032] In the formula, S lower is the target lower limit trajectory, S upper is the target upper limit trajectory; a min 、a maxare the set minimum acceleration and maximum acceleration values respectively; v0 is the initial velocity, and α is the acceleration trajectory to be planned; is an N-order diagonal matrix, where N is the prediction time domain; ψ = ones(N), is an N-order vector; and I is the unit matrix.
[0033] Preferably, the target lower limit trajectory S lower The minimum uniform speed V low The velocity trajectory through the intersection, where V low The calculation expression is:
[0034]
[0035] The target upper limit trajectory S upper The maximum uniform speed V upp The velocity trajectory through the intersection, where V upp The calculation expression is:
[0036]
[0037] In the formula, ΔL lgt is the distance between the vehicle and the traffic light at the intersection ahead; T g , T r , T c They are respectively the green light signal period of the traffic light ahead, the red light signal period of the traffic light ahead, and the current traffic light display time captured by the camera; sta_lgt is the traffic light status captured by the camera, including green and red status; V_lim is the speed limit value.
[0038] Preferably, the optimal control model based on the PMP minimum principle in step S3 has a specific mathematical expression as follows:
[0039]
[0040]
[0041] In the formula, is the fuel consumption, I bat is the battery current; x(t) is the state quantity, which is represented by the battery SOC here; u(t) is the control quantity, which is the current engine torque required by the vehicle; k is the weight coefficient of the input current; m f (t) is the fuel consumption; t0, t f are the starting time of the given prediction time domain; H is the corresponding Hamiltonian function; p(t) is the covariate variable, is the derivative of the battery SOC.
[0042] Preferably, in step S3, based on the target vehicle speed trajectory in the current prediction time domain obtained through planning, the required power at the current moment is calculated, and the specific mathematical expression is:
[0043] P t =V t ·(f·M·g·cosα+1 / 2·ρ·A·C D V t 2 +M·a t +M·g·sinα)·R wheel
[0044] Among them, P t is the power demand at the current moment, α is the road slope, f is the road friction coefficient, ρ is the air density, C D is the drag coefficient, R wheel is the wheel radius, g is the acceleration due to gravity, V t is the target vehicle speed trajectory obtained by optimization in step S2, A is the windward cross-sectional area of the vehicle, M is the mass of the vehicle, and a t The planned acceleration trajectory is optimized.
[0045] According to a second aspect of the present invention, a PHEV energy management system based on visual traffic information is provided, using any of the methods described above, the system comprising:
[0046] The upper perception module is used to extract effective traffic information from the collected image information using the YOLO image recognition model with pre-trained visual traffic information weights;
[0047] Speed planning module, used to predict speed trajectory based on valid traffic information;
[0048] The lower-level energy management module is used to predict the optimal control quantity based on the target vehicle speed trajectory obtained through planning and realize real-time energy management.
[0049] According to a third aspect of the present invention, there is provided an electronic device, comprising a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the program, any one of the methods described above is implemented.
[0050] According to a fourth aspect of the present invention, there is provided a computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, any one of the methods described above is implemented.
[0051] Compared with the prior art, the present invention has the following advantages:
[0052] 1) Based on the widespread application of vehicle-mounted cameras under the background of vehicle intelligence development, the present invention uses cameras to extract traffic information. Compared with the relative position information obtained in the past, more state quantity inputs can be obtained for the PHEV energy management system, including distance information and traffic sign information, which can effectively improve the energy management efficiency;
[0053] 2) The YOLO v5 framework used in the present invention to process the input of traffic information can ensure the real-time and reliability of the traffic information processing results, and provide good support for the real-time calculation of the energy management system;
[0054] 3) The QP quadratic programming strategy of the present invention performs speed trajectory planning within a given prediction time domain, and has better real-time performance;
[0055] 4) Taking battery current input as one of the cost indicators, the optimal control quantity is optimized and solved by using the optimal control model based on the PMP minimum principle, which theoretically guarantees the optimality and interpretability of optimization indicators such as vehicle longitudinal acceleration, fuel consumption, and battery life;
[0056] 5) The model predictive control (MPC) rolling optimization architecture is applied to solve the optimal control sequence of the engine's required torque at each forward iteration step, which increases the feasibility of real-time calculation. BRIEF DESCRIPTION OF THE DRAWINGS
[0057] Figure 1 It is a schematic diagram of the structure of the PHEV energy management system integrating visual traffic information of the present invention;
[0058] Figure 2 Training process for visual traffic information weights;
[0059] Figure 3 Online optimization process for real vehicle energy management. DETAILED DESCRIPTION
[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.
[0061] Example
[0062] First, a method embodiment of the present invention is given, a PHEV energy management method based on visual traffic information, the method comprising the following steps:
[0063] Step S1, obtaining image information in front of the target vehicle in real time through the on-board front-view camera, and extracting effective traffic information from the image information using the YOLO image recognition model with pre-trained visual traffic information weights; wherein the effective traffic information includes the longitudinal distance to the vehicle in front, the state and distance of the traffic light at the intersection in front, and the distance and speed limit value information of the speed limit sign in front;
[0064] Among them, the specific process of pre-training visual traffic information weights is:
[0065] 1) Modeling of conventional traffic environment and target vehicles in a 3D simulation environment;
[0066] The traffic environment includes but is not limited to: a standard lane of a certain length, lane lines, red and green traffic lights, speed limit signs, etc.; a forward-looking camera is installed directly in front of the front windshield of the target vehicle to capture the traffic scene ahead;
[0067] 2) Set a following vehicle scenario, with the speed trajectory of the front vehicle fixed and the rear vehicle being the target vehicle with a camera installed. Set the simulation step length to 10ms and perform a following vehicle simulation test under the set working conditions. During the simulation process, save the front view image taken by the front view camera in real time according to the simulation step length.
[0068] 3) Pre-calibrate the forward-looking simulation image and classify it according to the distance to the front vehicle, the distance to the front traffic light, and the distance to the front speed limit sign. Fit the polynomial relationship between the bounding-box side length and the three types of distances based on the distance: s1=f(L_fr), s2=f(L_lgt), s3=f(L_lim); where L_fr is the distance to the front vehicle, L_lgt is the distance to the traffic light, and L_lim is the distance to the front speed limit sign.
[0069] 4) Importing the classified images into the YOLO v5 image recognition model according to categories, and fitting neural network weight files of different categories through YOLO v5 image recognition training and offline iterative training; the neural network is divided into a first neural network, a second neural network and a third neural network according to categories; wherein the first neural network outputs the front vehicle distance L_fr, and the second neural network outputs the traffic light distance L_lgt, the traffic light state Sta_lgt and the traffic light display time T c , the third neural network outputs the speed limit sign distance L_lim and the speed limit value V_lim ahead.
[0070] Step S2: The extracted effective traffic information is used as the real-time state quantity and constraint quantity input, trajectory planning is performed within a given prediction time domain based on the QP quadratic programming strategy, and the target vehicle speed trajectory within the current prediction time domain is output; the mathematical expression of the QP quadratic programming model is:
[0071] 1) Quadratic programming objective function:
[0072]
[0073] In the formula, x is the optimization parameter, which refers to the acceleration trajectory to be planned; H and Q are coefficient matrices, and their expressions are:
[0074]
[0075]
[0076] Where M is the mass of the vehicle, A is the windward cross-sectional area of the vehicle, and C D is the drag coefficient, f is the road friction coefficient, and ρ is the air density; is an N-order diagonal matrix, N is the prediction time domain; ψ=ones(N), is an N-order vector;
[0077] 2) Constraints:
[0078]
[0079] ψ·v min ≤ψ·v0+ψ·a≤ψ·v max
[0080] ψ·a min ≤a≤ψ·a max
[0081] In the formula, S lower is the target lower limit trajectory, S upper is the target upper limit trajectory; a min 、a max are the set minimum acceleration and maximum acceleration values respectively; v0 is the initial velocity, and α is the acceleration trajectory to be planned; is an N-order diagonal matrix, N is the prediction time domain; ψ=ones(N), is an N-order vector; I is the unit matrix;
[0082] The target lower limit trajectory S lower The minimum uniform speed V low The velocity trajectory through the intersection, where V low The calculation expression is:
[0083]
[0084] The target upper limit trajectory S upper The maximum uniform speed V upp The velocity trajectory through the intersection, where V upp The calculation expression is:
[0085]
[0086] In the formula, ΔL lgt is the distance between the vehicle and the traffic light at the intersection ahead; T g 、T r 、T c They are respectively the green light signal period of the traffic light ahead, the red light signal period of the traffic light ahead, and the current traffic light display time captured by the camera; sta_lgt is the traffic light status captured by the camera, including green and red status; V_lim is the speed limit value.
[0087] Step S3, based on the target vehicle speed trajectory in the current prediction time domain obtained by planning, the required power at the current moment is calculated, and the battery current input is used as one of the cost indicators to construct an optimal control model based on the PMP minimum principle, and the optimal control amount of the engine required torque at the current moment is solved by Hamiltonian function optimization;
[0088] Among them, the mathematical expression of the required power at the current moment is:
[0089] P t =V t ·(f·M·g·cosα+1 / 2·ρ·A·C D V t 2 +M·a t +M·g·sinα)·R wheel
[0090] Among them, α is the road slope, f is the road friction coefficient, ρ is the air density, C D is the drag coefficient, R wheel is the wheel radius, g is the acceleration due to gravity, V t is the target vehicle speed trajectory obtained by optimization in step S2, A is the windward cross-sectional area of the vehicle, M is the mass of the vehicle, and a t The planned acceleration trajectory is optimized.
[0091] The optimal control model based on the PMP minimum principle has the following mathematical expression:
[0092]
[0093]
[0094] In the formula, is the fuel consumption, I bat is the battery current; x(t) is the state quantity, which is represented by the battery SOC here; u(t) is the control quantity, which is the current engine torque required by the vehicle; k is the weight coefficient of the input current; m f (t) is the fuel consumption; t0, t f are the starting time of the given prediction time domain; H is the corresponding Hamiltonian function; p(t) is the covariate variable, is the derivative of the battery SOC.
[0095] Step S4: adopt the model predictive control MPC rolling optimization architecture to solve the optimal control quantity sequence of the engine demand torque at each forward step of the iteration, and output the first control quantity in the solved control quantity sequence to the vehicle control unit.
[0096] Next, a system embodiment of the present invention is given, a PHEV energy management system based on visual traffic information, using any of the methods described above, the system comprising:
[0097] The upper perception module is used to extract valid traffic information from the collected image information using the YOLO image recognition model with pre-trained visual traffic information weights;
[0098] Speed planning module, used to predict speed trajectory based on valid traffic information;
[0099] The lower-level energy management module is used to predict the optimal control quantity based on the target vehicle speed trajectory obtained through planning and realize real-time energy management.
[0100] The electronic device of the present invention includes a central processing unit (CPU), which can perform various appropriate actions and processes according to computer program instructions stored in a read-only memory (ROM) or loaded from a storage unit into a random access memory (RAM). In the RAM, various programs and data required for device operation can also be stored. The CPU, ROM and RAM are connected to each other via a bus. An input / output (I / O) interface is also connected to the bus.
[0101] Multiple components in the device are connected to the I / O interface, including: input units, such as keyboards, mice, etc.; output units, such as various types of displays, speakers, etc.; storage units, such as disks, optical disks, etc.; and communication units, such as network cards, modems, wireless communication transceivers, etc. The communication unit allows the device to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunication networks.
[0102] The processing unit performs the various methods and processes described above, such as methods S1 to S4. For example, in some embodiments, methods S1 to S4 may be implemented as a computer software program, which is tangibly contained in a machine-readable medium, such as a storage unit. In some embodiments, part or all of the computer program may be loaded and / or installed on the device via a ROM and / or a communication unit. When the computer program is loaded into the RAM and executed by the CPU, one or more steps of methods S1 to S4 described above may be performed. Alternatively, in other embodiments, the CPU may be configured to execute methods S1 to S4 in any other appropriate manner (e.g., by means of firmware).
[0103] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chip systems (SOCs), load programmable logic devices (CPLDs), and the like.
[0104] The program code for implementing the method of the present invention can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer or other programmable data processing device, so that the program code, when executed by the processor or controller, enables the functions / operations specified in the flow chart and / or block diagram to be implemented. The program code can be executed entirely on the machine, partially on the machine, partially on the machine as a stand-alone software package and partially on a remote machine, or entirely on a remote machine or server.
[0105] In the context of the present invention, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
[0106] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed by the present invention, and these modifications or replacements should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.
Claims
1. A PHEV energy management method based on visual traffic information, characterized in that: The method comprises the following steps: Step S1, obtaining image information in front of the target vehicle in real time through the on-board front-view camera, and using the YOLO image recognition model with pre-trained visual traffic information weights to extract effective traffic information from the image information; Step S2: The extracted effective traffic information is used as the real-time state quantity and constraint quantity input, trajectory planning is performed within a given prediction time domain based on the QP quadratic programming strategy, and the target vehicle speed trajectory within the current prediction time domain is output; Step S3, based on the target vehicle speed trajectory in the current prediction time domain obtained by planning, the required power at the current moment is calculated, and the battery current input is used as one of the cost indicators to construct an optimal control model based on the PMP minimum principle, and the optimal control amount of the engine required torque at the current moment is solved by Hamiltonian function optimization; Step S4, using the model predictive control MPC rolling optimization architecture, solving the optimal control quantity sequence of the engine demand torque at each forward step of the iteration, and outputting the first control quantity in the solved control quantity sequence to the vehicle control unit; The mathematical expression of the QP quadratic programming model in step S2 is: 1) Quadratic programming objective function: Where x is the optimization parameter, which refers to the acceleration trajectory to be planned; H and Q are coefficient matrices, and their expressions are: Where M is the mass of the vehicle, A is the windward cross-sectional area of the vehicle, and C D is the drag coefficient, f is the road friction coefficient, and ρ is the air density; is an N-order diagonal matrix, where N is the prediction time domain; is an N-order vector; 2) Constraints: ψ·v min ≤ψ·v0+ψ·a≤ψ·v max ψ·a min ≤a≤ψ·a max In the formula, S lower is the target lower limit trajectory, S upper is the target upper limit trajectory; a min 、a max are the set minimum acceleration and maximum acceleration values respectively; v0 is the initial velocity, and α is the acceleration trajectory to be planned; is an N-order diagonal matrix, N is the prediction time domain; ψ=ones(N), is an N-order vector; I is the unit matrix; The target lower limit trajectory S lower The minimum uniform speed V low The velocity trajectory through the intersection, where V low The calculation expression is: The target upper limit trajectory S upper For the maximum uniform speed V upp The velocity trajectory through the intersection, where V upp The calculation expression is: Where, ΔL lgt is the distance between the vehicle and the traffic light at the intersection ahead; T g , T r , T c They are respectively the green light signal cycle of the traffic light ahead, the red light signal cycle of the traffic light ahead, and the current traffic light display time captured by the camera; sta_lgt is the traffic light status captured by the camera, including green and red status; V_lim is the speed limit value; The optimal control model based on the PMP minimum principle in step S3 is specifically expressed as follows: In the formula, I bat is the battery current; x(t) is the state quantity, which is represented by the battery SOC here; u(t) is the control quantity, which is the current engine torque required by the vehicle; k is the weight coefficient of the input current; m f (t) is the fuel consumption; t0, t f are the starting time of the given prediction time domain; H is the corresponding Hamiltonian function; p(t) is the covariate variable, is the differential of the battery SOC; In step S3, based on the target vehicle speed trajectory in the current prediction time domain obtained through planning, the required power at the current moment is calculated, and the specific mathematical expression is: P t =V t ·(f·M·g·cosα+1 / 2·ρ·A·C D V t 2 +M·a t +M·g·sinα)·R wheel Among them, P t is the power demand at the current moment, α is the road slope, f is the road friction coefficient, ρ is the air density, C D is the drag coefficient, R wheel is the wheel radius, g is the acceleration due to gravity, V t is the target vehicle speed trajectory obtained by optimization in step S2, A is the windward cross-sectional area of the vehicle, M is the mass of the vehicle, and a t The planned acceleration trajectory is optimized.
2. A PHEV energy management method based on visual traffic information according to claim 1, characterized in that: The effective traffic information in step S1 includes the longitudinal distance to the vehicle ahead, the state and distance of the traffic light at the intersection ahead, and the distance and speed limit value information of the speed limit sign ahead.
3. The PHEV energy management method based on visual traffic information according to claim 1, characterized in that: The YOLO image recognition model in step S1 that has pre-trained visual traffic information weights, and the visual traffic information weight pre-training process is: 1) Modeling the conventional traffic environment and the target vehicle in a three-dimensional simulation environment; wherein a forward-looking camera is installed directly in front of the front windshield of the target vehicle to capture the traffic scene ahead; 2) Set a vehicle-following scenario, with the speed trajectory of the front vehicle fixed and the rear vehicle being the target vehicle with a camera installed. Set the simulation step length and perform a vehicle-following simulation test under the set working conditions. During the simulation process, save the front view image taken by the front view camera in real time according to the simulation step length. 3) Pre-calibrate the forward-looking simulation image and classify it based on the distance to the vehicle ahead, the distance to the traffic light ahead, and the distance to the speed limit sign ahead; 4) Importing the classified images into the YOLO v5 image recognition model according to categories, and fitting neural network weight files of different categories through YOLO v5 image recognition training and offline iterative training; the neural network is divided into a first neural network, a second neural network and a third neural network according to categories; wherein the first neural network outputs the front vehicle distance L_fr, the second neural network outputs the traffic light distance L_lgt and the traffic light state Sta_lgt, and the third neural network outputs the front speed limit sign distance L_lim and the speed limit value V_lim.
4. A PHEV energy management system based on visual traffic information, characterized in that: The method according to any one of claims 1 to 3 is adopted, wherein the system comprises: The upper perception module is used to extract valid traffic information from the collected image information using the YOLO image recognition model with pre-trained visual traffic information weights; Speed planning module, used to predict speed trajectory based on valid traffic information; The lower-level energy management module is used to predict the optimal control quantity based on the target vehicle speed trajectory obtained through planning and realize real-time energy management.
5. An electronic device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the program, the method according to any one of claims 1 to 3 is implemented.
6. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 3 is implemented.
