Intelligent network connection electric vehicle multi-objective optimization control method based on data driving and application thereof

Abstract

The invention relates to a multi-objective optimization control method for an intelligent networked electric vehicle based on data driving and an application thereof. The method comprises the following steps: extracting a vehicle model for a single intelligent networked electric vehicle and obtaining a state representation of the vehicle model in a vehicle queue; constructing a data-driven prediction model based on a subspace identification method and historical data, and constructing a linear input and output mapping relation; constructing a multi-objective optimization function, and executing optimization control based on a multi-objective optimization result; the method is applied to a controller of an intelligent network-connected electric vehicle in a vehicle queue, the controller collects acceleration and braking operation data to track torque and outputs control quantity for executing optimization control, and queue control is achieved. According to the method, the problem of control performance degradation caused by modeling errors or system complexity is effectively solved; a multi-objective optimization problem is introduced, so that the practicability and robustness of the system under a complex working condition are enhanced; the control performance is balanced while the prediction precision is kept, and the operation safety, the energy saving performance and the cooperation efficiency of the intelligent network connection electric vehicle queue are effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of traffic control systems, in particular to a data-driven intelligent connected electric vehicle multi-objective optimization control method and application thereof. BACKGROUND

[0002] In recent years, intelligent connected electric vehicles, as an important embodiment of the intelligent and information development of the transportation system, are gradually leading the deep transformation of the automotive industry. By carrying advanced sensors, communication modules and automatic control systems, intelligent connected electric vehicles can realize efficient information interaction between vehicles, roads and even the cloud, thereby enhancing driving safety, improving road traffic efficiency, and laying a technical foundation for emerging applications such as vehicle fleet coordination control and autonomous driving. Under this background, the research on modeling and control methods for multi-vehicle coordination has become a key issue to ensure system performance and operational safety.

[0003] Traditional control methods generally rely on accurate physical modeling and parameter identification processes. However, due to the complexity of electric vehicle structure and the high dynamic nature of its operating environment, modeling methods based on prior models have the bottleneck of being unable to accurately capture the true behavior of the system, especially when facing battery state changes, load disturbances and multi-source information interaction, which easily affects the control performance.

[0004] Chinese patent CN119087808A discloses an intelligent connected vehicle distributed optimization control method based on reinforcement learning strategy, which has a hierarchical structure and includes the following steps: establishing a vehicle longitudinal dynamics model and setting control objectives for the upper and lower layers; designing the upper layer, i.e., the trajectory optimization control layer; a trajectory planning controller based on a distributed convex optimization algorithm combined with an internal model principle, a controller designed based on the internal model principle to remove external disturbances in the upper trajectory planning process; designing the lower layer, i.e., the tracking control layer; an optimal trajectory tracking controller based on the Actor-Critic framework of reinforcement learning; analyzing the stability and convergence of the tracking control system through Lyapunov stability criterion to ensure accurate tracking of the reference trajectory; and verifying the feasibility of the proposed algorithm through simulation experiments. Although this application studies the trajectory planning and trajectory tracking control problems in the connected vehicle platoon control, it breaks away from the dependence on precise vehicle dynamics equations, but in the trend of increasingly complex control objectives, the optimization of a single performance indicator is difficult to meet the actual operational requirements. The control of intelligent connected electric vehicles not only needs to ensure the safe distance between the platoon and the dynamic stability of the vehicle, but also needs to consider driving comfort and energy economy. Based on this, building a data-driven model predictive control method that takes into account multiple objectives will become a research hotspot. SUMMARY

[0005] The application solves the problems in the prior art and provides a data-driven intelligent networked electric vehicle multi-objective optimization control method and application thereof, which realizes prediction control of the intelligent networked electric vehicle based on a multi-objective data-driven model, is suitable for a coordinated control system of an electric vehicle queue under a complex power structure, deals with the uncertainty of a complex vehicle power system model, and provides a safe, comfortable and economical driving strategy.

[0006] The technical scheme adopted by the application is a data-driven intelligent networked electric vehicle multi-objective optimization control method, which comprises the following steps:

[0007] S1 extracts a vehicle model for a single intelligent networked electric vehicle and obtains a state representation of the vehicle in a vehicle queue;

[0008] S2 constructs a data-driven prediction model based on a subspace identification method and historical data, and then constructs a linear input-output mapping relationship;

[0009] S3 constructs a multi-objective optimization function;

[0010] S4 performs optimization control based on the obtained multi-objective optimization result.

[0011] Preferably, in S1, the vehicle model comprises a vehicle kinematics model,

[0012]

[0013] wherein, represents a driving torque, F represents a driving force, R is a wheel radius, m is a vehicle mass, S is a displacement, and V represents a speed, is an acceleration;

[0014] The vehicle model further comprises an SOC state,

[0015]

[0016] wherein, is a battery capacity of the current intelligent networked electric vehicle, is an initial state of the, and I is a current generated by a motor of the current intelligent networked electric vehicle.

[0017] Preferably, in the vehicle queue, the state of the i-th vehicle at time k is , the control input is , is a control output, and the state representation satisfies,

[0018]

[0019] wherein, is a state transition matrix, is an input matrix, is an output matrix;

[0020] associated with the speed variation, displacement variation and SOC variation of the current vehicle.

[0021] Preferably, in S2, the historical input data and output data of any vehicle in a period of time are organized as a Hankel matrix with time sequence characteristics, a least square optimization problem is constructed, a subspace linear mapping matrix for prediction is estimated, and a data-driven prediction model between the input and future state is established.

[0022] Preferably, based on the data-driven prediction model, the linear input-output mapping relationship is constructed to satisfy,

[0023]

[0024] wherein, is a subspace prediction matrix corresponding to the historical input and output trajectory, is a subspace prediction matrix related to the future input trajectory, W i p ( k ) = [ Y i p ( k ) U i p ( k ) ] ; here p and f respectively represent present and future.

[0025] Preferably, in S3, the multi-objective optimization function is associated with the prediction variables of driving safety, driving comfort and energy economy.

[0026] Preferably, the total cost function is,

[0027]

[0028] wherein, represents a weight coefficient corresponding to each optimization objective, in which, is respectively correspond to the cost functions of driving safety, driving comfort and energy economy, is a transmission ratio, τ is an index of N time steps from time k.

[0029] Preferably, the multi-objective optimization function further includes a soft constraint condition set in conjunction with the total cost function, satisfying

[0030]

[0031] wherein, and are respectively the maximum and minimum limit values of the output, and are respectively the maximum and minimum limit values of the control.

[0032] Preferably, singular value decomposition is introduced to the subspace prediction matrix, and the feedforward control law of the data-driven model predictive control is mapped to a low-dimensional feature subspace, and the first component is used to perform optimization control.

[0033] The application of the data-driven intelligent connected electric vehicle multi-objective optimization control method is applied to the controller of the intelligent connected electric vehicle in the vehicle queue, the controller collects acceleration and braking operation data to track torque, outputs control variables for performing optimization control, and realizes queue control.

[0034] The application relates to a data-driven intelligent connected electric vehicle multi-objective optimization control method and application thereof, the method extracts a vehicle model for a single intelligent connected electric vehicle and obtains a state representation of the vehicle in a vehicle queue; a data-driven prediction model is constructed based on a subspace identification method and historical data, and a linear input-output mapping relationship is constructed; a multi-objective optimization function is constructed, and based on the obtained multi-objective optimization result, optimization control is performed; the method is applied to the controller of the intelligent connected electric vehicle in the vehicle queue, the controller collects acceleration and braking operation data to track torque, outputs control variables for performing optimization control, and realizes queue control.

[0035] The application has the following advantages:

[0036] (1) By constructing an agent model using input and output data, the dependence on accurate knowledge of system dynamics in traditional mechanism modeling is avoided, modeling and prediction can be realized without an explicit physical model of the system, the efficiency and robustness of model construction are significantly improved, a Hankel matrix can be constructed directly based on historical operation data, and system dynamic characteristics are extracted through a subspace identification algorithm, so that effective prediction of future system behavior is realized, and the problem of control performance degradation caused by modeling errors or system complexity is effectively overcome;

[0037] (2) A multi-objective optimization problem is introduced in the control strategy, the requirements of driving safety, energy economy and driving comfort are comprehensively considered, a weighted cost function is constructed and a flexible constraint mechanism is embedded, the adaptability of the control algorithm to diversified targets in actual driving scenarios is improved, the contradiction between different control targets can be dynamically balanced, the control strategy can be flexibly adjusted under the premise of ensuring constraint feasibility, and the practicability and robustness of the system under complex working conditions are enhanced.

[0038] (3) It is particularly suitable for complex and nonlinear intelligent connected electric vehicle systems, and can balance control performance while maintaining prediction accuracy, effectively improving the safety, energy saving and coordination efficiency of intelligent connected electric vehicle queue operation. BRIEF DESCRIPTION OF DRAWINGS

[0039] Figure 1A flow chart of the method of the present application;

[0040] Figure 2 A vehicle platoon communication topology structure;

[0041] Figure 3 Figures (a) and (b) are vehicle platoon spacing results for the 2nd follower vehicle (F2) and the 3rd follower vehicle (F3), respectively;

[0042] Figure 4 Figures (a) and (b) are vehicle platoon speed results for the 2nd follower vehicle (F2) and the 3rd follower vehicle (F3), respectively;

[0043] Figure 5 Figures (a) and (b) are vehicle platoon SOC results for the 2nd follower vehicle (F2) and the 3rd follower vehicle (F3), respectively. DETAILED DESCRIPTION

[0044] The present application will be further described in conjunction with the embodiments below, but the scope of protection of the present application is not limited thereto.

[0045] The present application relates to a data-driven intelligent connected electric vehicle multi-objective optimization control method, in the embodiment, a vehicle platoon coordination simulation platform experiment platform is constructed, Matlab and IPG-CarMaker software are fused, and the platform runs on a computer, the computer is configured as an Intel i7-14700KF (3.4GHz) central processing unit and an RTX4060 graphics processing unit, and fully meets the calculation requirements of actual vehicle application; the power system of the intelligent connected electric vehicle is modeled in detail in the IPG-CarMaker, including chassis, battery, motor, clutch, engine and tire model; the data-driven prediction model proposed by the method is deployed on the host computer based on the Matlab language, and is used to generate an optimal control sequence in real time; considering that torque cannot be directly used to drive the vehicle, a PID-based bottom controller is designed, so that each intelligent connected electric vehicle can track the torque through acceleration and braking operations, and then realize platoon control; the algorithm is compiled as C++ code and embedded into the IPG-CarMaker software for running, real-time information interaction between vehicles is realized through the SimNet module of the IPG-CarMaker, the SimNet module supports vehicle connection between multiple independent CarMaker instances, and a closed-loop control structure is constructed.

[0046] The embodiment is configured to drive three intelligent connected electric vehicles along a straight path under preset conditions, which meets the basic platoon test requirements, and the preset conditions include:

[0047] 1) the vehicle drives on a dry and flat road, and the longitudinal slip of the tire can be ignored;

[0048] 2) The body structure is symmetrical and has good rigidity;

[0049] 3) The yaw and pitch movements of the vehicle are ignored;

[0050] 4) The effects of the internal resistance, temperature change and capacity attenuation of the battery are ignored;

[0051] The lead vehicle has optimal speed, spacing and SOC, which are regarded as reference states, and a reference speed signal of the lead vehicle is derived from a preset FTP75 standard operating condition block in Simulink, which can simulate variable driving conditions such as frequent idling, starting and stopping and acceleration in urban traffic. According to the prior art, the general headway between vehicles should be 2-3 seconds to ensure sufficient response time to avoid collision.

[0052] The method comprises the following steps:

[0053] (1) extracting a vehicle model for a single intelligent connected electric vehicle and obtaining a state representation of the vehicle in a vehicle queue;

[0054] (2) constructing a data-driven prediction model based on a subspace identification method and historical data, and then constructing a linear input-output mapping relationship;

[0055] (3) constructing a multi-objective optimization function;

[0056] (4) performing optimization control based on the obtained multi-objective optimization result.

[0057] The following will be described in detail in combination with the steps. Unless otherwise specified, the vehicle mentioned is an intelligent connected electric vehicle.

[0058] (1) extracting a vehicle model for a single intelligent connected electric vehicle and obtaining a state representation of the vehicle in a vehicle queue;

[0059] The vehicle model includes a vehicle kinematics model,

[0060]

[0061] wherein, represents the driving torque, F represents the driving force, R is the wheel radius, m is the vehicle mass, S is the displacement, V represents the speed, is the acceleration;

[0062] In the vehicle operation, the actual driving torque of the intelligent connected electric vehicle can be represented as

[0063]

[0064] wherein, is the driving torque output by the transmission, is the required motor output torque, is the transmission ratio, is the transmission system efficiency;

[0065] The current I generated by the motor of the intelligent connected electric vehicle can be expressed as:

[0066]

[0067] wherein, is the back electromotive force coefficient, is the motor speed and voltage, respectively;

[0068] The vehicle model also includes the SOC state, i.e., the state of charge (SOC) of the battery,

[0069]

[0070] wherein, is the current battery capacity of the intelligent connected electric vehicle, is the initial state of , and I is the current generated by the motor of the intelligent connected electric vehicle.

[0071] In the vehicle queue, the state of the i-th vehicle (i is a positive integer) at time k is i ∈ ( 1 , M ] , and the control input is , , The error state model of the i-th vehicle in the intelligent connected electric vehicle queue is constructed using Taylor series expansion, and the state is represented as

[0072]

[0073] wherein, is the state transition matrix, A i ( k ) = [ 1 0 0 0 T s + 1 0 0 0 1 ] , T s is the sampling time;

[0074] is the input matrix, B i ( k ) = [ T e , i η i ξ i T s / ( m i R i ) 0 − T s / ( K e C b a t ) ] ;

[0075] is the output matrix;

[0076] is associated with the speed change, displacement change and SOC change of the current vehicle, and in this embodiment, X i ( k ) = [ V ˜ i ( k ) S ˜ i ( k ) S O C ̃ i ( k ) ] = [ V i ( k ) − V 1 ( k ) S i ( k ) − S 1 ( k ) S O C i ( k ) − S O C 1 ( k ) ] ;

[0077] U i ( k ) = [ T ˜ e , i ( k ) ] = [ T e , i ( k ) − T e , 1 ( k ) ] ​ .

[0078] (2) Based on the subspace identification method and historical data to construct a data-driven prediction model, and then construct a linear input-output mapping relationship;

[0079] The historical input data and output data of any vehicle in a period of time are organized into a Hankel matrix with time sequence characteristics, which can fully reflect the dynamic evolution law of the system in structure;

[0080] Construct a least squares optimization problem to estimate the subspace linear mapping matrix for prediction, and establish a data-driven prediction model between input and future state;

[0081] This step does not depend on the prior physical model of the system, but directly extracts the dynamic characteristics of the system from the observation data, has good data-driven characteristics, effectively alleviates the control problems caused by modeling errors, and can be directly applied to the prediction process without identifying the local state space model; Specifically, the equation constructed by the subspace-based linear predictor in the form of Hankel matrix is,

[0082]

[0083] Among them, Y i p ( k ) = [ Y i ( k ) ⋯ Y i ( k + N − L ) ⋮ ⋱ ⋮ Y i ( k + L − 1 ) ⋯ Y i ( k + N − 1 ) ] , Y i f ( k ) = [ Y i ( k + L ) ⋯ Y i ( k + N ) ⋮ ⋱ ⋮ Y i ( k + 2 L − 1 ) ⋯ Y i ( k + L + N − 1 ) ] , U i p ( k ) = [ U i ( k ) ⋯ U i ( k + N − L ) ⋮ ⋱ ⋮ U i ( k + L − 1 ) ⋯ U i ( k + N − 1 ) ] , U i f ( k ) = [ U i ( k + L ) ⋯ U i ( k + N ) ⋮ ⋱ ⋮ U i ( k + 2 L − 1 ) ⋯ U i ( k + L + N − 1 ) ] , X i p ( k ) = [ X i ( k ) ⋮ X i ( k + N − L ) ] , X i f ( k ) = [ X i ( k + L ) ⋮ X i ( k + N ) ] , Γ i ( k ) = [ C i ( k ) A i 1 − 1 ⋮ C i ( k ) A i L − 1 ] , ϒ i ( k ) = [ A i L − 1 B i ⋮ A i L − L B i ] , H i ( k ) = [ C i A i L − L B i ⋯ C i A i L − 1 B i ⋮ ⋱ ⋮ 0 ⋯ 0 ] ;

[0084] The above matrix is the system related state matrix, N is the prediction step, and L is the data excitation dimension.

[0085] Based on the data-driven prediction model, the linear input-output mapping relationship is constructed, which can obtain

[0086]

[0087] In the formula, Moore-Penrose pseudo-inverse matrix; Then substitute the above formula into the Hankel matrix , get,

[0088]

[0089] The linear predictor equation is derived.

[0090]

[0091] in, This is the subspace prediction matrix corresponding to the historical input and output trajectories. This is the subspace prediction matrix associated with the future input trajectory. W i p ( k ) = [ Y i p ( k ) U i p ( k ) ] Using the orthogonal projection property of QR decomposition, the solution can be obtained as follows:

[0092] [ W i p ( k ) U i f ( k ) ] = [ R i , 1 1 ( k ) 0 R i , 2 1 ( k ) R i , 2 2 ( k ) ] [ Q i , 1 ( k ) Q i , 2 ( k ) ]

[0093] Orthogonal projection provides an effective solution to the aforementioned least squares problem, which can be expressed as follows:

[0094] Y ^ i ( k ) = Y i f ( k ) [ W i p ( k ) U i f ( k ) ] † [ W i p ( k ) U i f ( k ) ] = [ R i , 3 1 ( k ) R i , 3 2 ( k ) ] T [ R i , 1 1 ( k ) 0 R i , 2 1 ( k ) R i , 2 2 ( k ) ] † [ W i p ( k ) U i f ( k ) ] = [ L w , i ( k ) L u , i ( k ) ] [ W i p ( k ) U i f ( k ) ] T

[0095] The predicted input / output sequence can be calculated using the above method, and this sequence will then be used to design a data-driven model predictive control strategy.

[0096] (3) Construct a multi-objective optimization function;

[0097] The multi-objective optimization function is associated with the predictor variables of driving safety, driving comfort and energy economy; vehicle speed and control torque are key indicators for measuring driving comfort, while vehicle spacing and battery state of charge are core factors for ensuring driving safety and energy economy. It should be noted that the predicted sequence of the preceding vehicle i−1 is regarded as the optimal tracking target of the current vehicle i. In terms of economic objectives, the aim is to keep the SOC at a high level and as close as possible to the SOC value of the reference state.

[0098] The total cost function is,

[0099]

[0100] in, This represents the weight coefficient corresponding to each optimization objective. Corresponding to driving safety ( ), driving comfort ) and energy economy ( The cost function of )

[0101]

[0102]

[0103]

[0104] Q and P are weight coefficients of vehicle speed and input torque respectively; N represents the prediction step, 、 and represent the distance error, speed error and state of charge error respectively, represent the vehicle acceleration.

[0105] is the transmission ratio, and τ is the index of N time steps from time k.

[0106] The distributed model predictive control algorithm has a significant advantage in effectively processing the operating constraints of the cost function, and based on the driver-centered driving strategy, the multi-objective optimization function also includes soft constraint conditions set in cooperation with the total cost function, meeting

[0107]

[0108] wherein, and are the maximum and minimum limit values of the output respectively, and are the maximum and minimum limit values of the control respectively.

[0109] (4) Based on the obtained multi-objective optimization result, performing optimization control;

[0110] The singular value decomposition is introduced to the subspace prediction matrix, the feedforward control law of the data-driven model predictive control is mapped to a low-dimensional feature subspace, and the first component is used to perform optimization control.

[0111] This operation mode maps the original high-dimensional optimization problem to a lower-dimensional feature subspace, thereby realizing efficient calculation and structural reconstruction of the control gain. Under this framework, the derivation of the control law not only considers the trade-off between multiple performance indicators, but also significantly reduces the solution dimension and computational complexity, enhancing the implementability and response efficiency of the model in real-time control.

[0112] The feedforward control law of the data-driven model predictive control can be described as

[0113]

[0114] Let the vector L i ( k ) = [ L w , i ( k ) L u , i ( k ) ] , and the singular value decomposition method is introduced to decompose L i :

[0115]

[0116] wherein, and is a unitary matrix, is a diagonal matrix; substituting and solving can obtain:

[0117]

[0118] The first control component among them is used to guide the operation of the vehicle fleet.

[0119] The application also relates to an application of the data-driven intelligent networked electric vehicle multi-objective optimization control method.

[0120] Three groups of comparative experiments including model predictive control (MPC), PID and the proposed data-driven model predictive control (DMPC) are set up, as shown in Figures 3~5 The proposed DMPC algorithm can realize the safe, energy-saving and comfortable driving conditions of the intelligent networked electric vehicle.

[0121] Those skilled in the art will understand that embodiments of the present application can be provided as methods, systems, or computer program products. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage, etc.) containing computer-usable program code.

[0122] The present application is described with reference to flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, as well as combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing apparatus to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing apparatus produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The functions specified in one or more flows and / or blocks

[0123] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0124] The computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions that are executed on the computer or other programmable apparatus provide steps for implementing the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0125] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the embodiments by those of skill in the art once they have the benefit of the present disclosure. Therefore, the appended claims are intended to encompass within their scope all possible variations and modifications of the preferred embodiments. 1

[0126] It is apparent that a person skilled in the art can make various changes and modifications to the application without departing from the spirit and scope thereof. Therefore, if these modifications and changes fall within the scope of the claims and their equivalents, it is intended to include them in the application.

Claims

1. A data-driven multi-objective optimization control method for intelligent connected electric vehicles, characterized in that: The method includes the following steps: S1 extracts the vehicle model for a single intelligent connected electric vehicle and obtains its state representation in the vehicle queue. S2 constructs a data-driven prediction model based on subspace identification methods and historical data, and then constructs a linear input-output mapping relationship; S3 constructs a multi-objective optimization function; S4 performs optimization control based on the obtained multi-objective optimization results.

2. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 1, characterized in that: In S1, the vehicle model includes the vehicle kinematics model. ， in, The driving torque is represented by F, the driving force by R, the wheel radius by m, the vehicle mass by S, and the velocity by V. For acceleration; The vehicle model also includes SOC status. ， in, For the current battery capacity of intelligent connected electric vehicles, for In the initial state, I represents the current generated by the motor of the current intelligent connected electric vehicle.

3. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 2, characterized in that: In the vehicle queue, the state of vehicle i at time k is: The control input is , To control the output, the state representation must satisfy, ， in, Here is the state transition matrix. For the input matrix, This is the output matrix; It is related to the current changes in vehicle speed, displacement, and SOC.

4. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 1, characterized in that: In S2, the historical input and output data of any vehicle over a period of time are organized into a Hankel matrix with temporal characteristics. A least squares optimization problem is constructed to estimate the subspace linear mapping matrix used for prediction and establish a data-driven prediction model between the input and the future state.

5. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 4, characterized in that: Based on the data-driven prediction model, the linear input-output mapping relationship is constructed, satisfying the following: ， in, This is the subspace prediction matrix corresponding to the historical input and output trajectories. This is the subspace prediction matrix associated with the future input trajectory. .

6. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 1, characterized in that: In S3, the multi-objective optimization function is associated with the predictors of driving safety, driving comfort, and energy economy.

7. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 6, characterized in that: The total cost function is, ， in, This represents the weight coefficient corresponding to each optimization objective. middle, for The time corresponds to the cost functions for driving safety, driving comfort, and energy economy, respectively. Let τ be the transmission ratio, and τ be the index of the N time steps starting from time k.

8. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 7, characterized in that: The multi-objective optimization function also includes soft constraints set in conjunction with the total cost function, satisfying... ， in, and Output the maximum and minimum limit values ​​respectively. and These control the maximum and minimum limits, respectively.

9. The data-driven multi-objective optimization control method for intelligent connected electric vehicles according to claim 5, characterized in that: Singular value decomposition is introduced into the subspace prediction matrix to map the feedforward control law of data-driven model predictive control to a low-dimensional feature subspace, and the first component is used to perform optimal control.

10. An application of the data-driven multi-objective optimization control method for intelligent connected electric vehicles as described in any one of claims 1 to 9, characterized in that: The controller for intelligent connected electric vehicles used in vehicle platoons collects acceleration and braking operation data to track torque, outputs control quantities to perform optimized control, and realizes platoon control.

Images