A new energy vehicle adaptive energy management method based on multi-source information fusion

Abstract

The application discloses a new energy automobile adaptive energy management method based on multi-source information fusion, relates to the technical field of vehicle energy management, and solves the problem that the comfort of cabin members and the kinetic energy recovery efficiency of an existing new energy automobile kinetic energy recovery system cannot be considered simultaneously due to fixed gears. The method comprises the following steps: collecting multi-source data in the process of vehicle driving, processing the multi-source data into a system state vector x(k); identifying a driving scene and a driving intention according to the system state vector x(k), and calculating an expected deceleration a des (k) and a target kinetic energy recovery intensity coefficient λ(k) according to the system state vector x(k); outputting motor recovery torque T reg (k) according to the driving scene, the driving intention, the expected deceleration a des (k) and the target kinetic energy recovery intensity coefficient λ(k); and adjusting the kinetic energy recovery intensity according to the motor recovery torque T reg (k). The application solves the contradiction between driving comfort and kinetic energy recovery efficiency caused by fixed adjustment of a small amount of kinetic energy recovery intensity gears.

Description

Technical Field

[0001] This invention relates to the field of vehicle energy management technology, and more specifically, to an adaptive energy management method for new energy vehicles based on multi-source information fusion. Background Technology

[0002] Today, new energy vehicles are gradually becoming more common in private households. While enjoying clean energy, green travel, and lower travel costs, people are also placing higher demands on the passenger experience of new energy vehicles. Therefore, how to improve kinetic energy recovery efficiency while maximizing passenger comfort has become a key research focus for major automakers and is one of the core pain points in the energy management of new energy vehicles. The following are relevant patents related to energy management in new energy vehicles:

[0003] Among them, patent CN110588656B discloses an adaptive kinetic energy recovery method and system based on road and road condition information. Its core is a data acquisition module that obtains map and road condition information. When the driver releases the accelerator without applying the brakes, a prediction module predicts the future path, a judgment module determines whether deceleration is necessary, a processing module determines and updates the optimal deceleration strategy, and the kinetic energy recovery module executes the recovery. This eliminates the need for manual settings by the driver, reducing operational stress, improving ride comfort and energy efficiency, and increasing vehicle range.

[0004] Patent CN108058615B discloses a method and device for vehicle braking energy recovery. The core of this patent is that the first vehicle acquires surrounding vehicle driving information and preset road information, combines this with its own driving speed to determine the target torque, and then identifies the driving intention through pedal and control status to determine the vehicle's required torque and control the motor to recover energy. This method considers both internal and external vehicle information, avoids passive recovery, and effectively improves the braking energy recovery rate.

[0005] Patent CN112824130B discloses BAIC's braking energy recovery level control method, device, and vehicle, which is applied to the vehicle control unit (VCU). The core of this patent is to acquire historical recovery level information from the user's most recent power outage and send it to the MCU when the vehicle switches to D gear. It then acquires driving condition information from the user within a preset time period (including accelerator / brake pedal opening, vehicle speed, steering wheel turning angle change rate, etc.), and, combined with the power battery SOC, determines the first recovery level among three levels (comfort, strong, and strong) and sends it to the MCU, while simultaneously saving this level for future use. This achieves intelligent switching, simplifies operation, reduces gear buttons and wiring harnesses, lowers costs, and improves the user experience.

[0006] The above application documents reveal the following problems with the existing technology: 1. Existing kinetic energy recovery intensity adjustments are mostly fixed in two to three levels, lacking flexibility; 2. When the kinetic energy recovery intensity is at a high level, releasing the accelerator will cause discomfort to passengers due to sudden deceleration; when the kinetic energy recovery intensity is at a low level, the kinetic energy recovery efficiency is greatly reduced. 3. Having only a single kinetic energy recovery strategy for different operating conditions not only affects the passenger experience but also hinders the improvement of kinetic energy recovery efficiency. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to provide an adaptive energy management method for new energy vehicles based on multi-source information fusion, which addresses the shortcomings of existing technologies and solves the problem that existing new energy vehicle kinetic energy recovery systems cannot simultaneously achieve both passenger comfort and kinetic energy recovery efficiency due to fixed gear positions.

[0008] The present invention discloses an adaptive energy management method for new energy vehicles based on multi-source information fusion. The method includes: collecting multi-source data during vehicle operation; processing the multi-source data into a system state vector x(k); identifying the driving scenario and driving intention based on the system state vector x(k); and simultaneously calculating the desired deceleration a based on the system state vector x(k). des (k) and the target kinetic energy recovery intensity coefficient λ(k); based on the driving scenario, driving intention, and desired deceleration a des (k) and the target kinetic energy recovery intensity coefficient λ(k) output motor recovery torque T reg (k); based on the motor recovery torque T reg (k) Adjust the intensity of kinetic energy recovery.

[0009] To further improve the process, the desired deceleration a is calculated based on the system state vector x(k). des The method for determining the system state vector x(k) and the target kinetic energy recovery intensity coefficient λ(k) is as follows: the system state vector x(k) includes the road slope vector γ(k) and the throttle opening vector θ(k). acc (k), Braking opening vector θ brk (k), the distance vector df1(k) to the vehicle in front during driving, the distance vector df2(k) to the vehicle behind during driving, and the relative velocity vector Δv(k) to the vehicle in front, based on the road slope vector γ(k) and the ignition switch opening vector θ acc (k), Braking opening vector θ brk (k) Distance vector d between the vehicle in front during driving f1 (k) Distance vector d between the vehicle and the following vehicle during the driving process f2 The desired deceleration a is calculated from (k) and the relative velocity vector Δv(k) with the vehicle in front. des (k); The method for calculating the target kinetic energy recovery intensity coefficient λ(k) based on the system state vector x(k) is as follows: the system state vector x(k) further includes the battery SOC vector SOC(k), road slope vector γ(k), vehicle speed vector v(k), acceleration vector a(k), and driver personalized preference parameter mode(k). The energy recovery priority index coefficient is calculated based on the battery SOC vector SOC(k), road slope vector γ(k), vehicle speed vector v(k), acceleration vector a(k), and driver personalized preference parameter mode(k). β(k) Set the foundation strength λ user (k), based on the aforementioned basic strength λ user (k) Energy recovery priority index coefficient β(k) and expected deceleration a des The target kinetic energy recovery intensity coefficient λ(k) is calculated.

[0010] Furthermore, the desired deceleration a des The calculation expression is, ; Where k1 is the calibration coefficient of the first expected deceleration calculation model, k2 is the calibration coefficient of the second expected deceleration calculation model, k3 is the calibration coefficient of the third expected deceleration calculation model, and k4 is the calibration coefficient of the fourth expected deceleration calculation model. a des (k) is the desired deceleration vector. c (k) is the road slope vector. v (k) is the vehicle speed vector. i acc (k) is the gate opening vector, i brk (k) is the braking opening vector. v tar To combine the road slope vector γ(k) and the throttle opening vector θ acc (k), Braking opening vector θ brk (k) Distance vector d between the vehicle in front during driving f1 (k) Distance vector d between the vehicle and the following vehicle during the driving process f2 The target vehicle speed is obtained from (k) and the relative velocity vector Δv(k) with the vehicle in front.

[0011] Furthermore, the energy recovery priority index coefficient β(k) The calculation expression is, ; in, m SOC (·) This is the normalized membership function that reflects the battery state. m γ (·) The normalized membership function reflects the road slope. m cvcle (·) To reflect the normalized membership function of vehicle operating conditions, m νref (·) To provide a normalized membership function that reflects the driver's individual preferences, ψ SOC As the first weight parameter, ψ γ This is the second weighting parameter. ψ cvcle As the third weighting parameter, ψ νref This is the fourth weighting parameter. β(k) Here, SOC(k) is the energy recovery priority index coefficient, and SOC(k) is the battery SOC vector. v (k) is the vehicle speed vector. a (k) is the acceleration vector, and mode(k) is the driver's personalized preference parameter.

[0012] Furthermore, the calculation expression for the target kinetic energy recovery intensity coefficient λ(k) is as follows: ; in, β(k) This is the energy recovery priority index coefficient. a des (k) is the desired deceleration vector. K β The first adjustable gain, K α For the second adjustment gain, β 0 is the first reference point for calibration. α 0 is the second reference point for calibration. λ(k) is the saturation function, and λ(k) is the target kinetic energy recovery intensity coefficient.

[0013] Furthermore, based on the aforementioned driving scenario, driving intention, and desired deceleration a des (k) and the target kinetic energy recovery intensity coefficient λ(k) output motor recovery torque command T reg The method for (k) is as follows: Based on the jerk j(k) and acceleration a(k), based on the desired deceleration a des (k) The optimization objective function is calculated. J opt Through the optimization objective function J optThe recovery intensity coefficient λ(k) is initially optimized, and then further adjusted based on the driving scenario, driving intention, and acceleration j(k) to generate the motor recovery torque command T. reg (k).

[0014] Furthermore, the optimization objective function J opt The calculation expression is, ; in, J opt To optimize the objective function, a(k+i) For the predicted acceleration, j(k+i) For jerk, ξ1 is the first calibration weighting coefficient, ξ2 is the second calibration weighting coefficient, and ξ3 is the third calibration weighting coefficient. P gen (k+i) The power output of the motor. a des (k+i) is the desired deceleration vector. N To predict the length of the time domain.

[0015] Furthermore, based on the driving scenario, driving intention, and acceleration j(k), the initially optimized recovery intensity coefficient λ(k) is further adjusted to form the motor recovery torque T. reg The method for (k) is as follows: When the driving scenario is a vehicle going downhill or traveling at high speed and the acceleration j is within the preset acceleration tolerance range, the recovery intensity coefficient λ(k) is set to the preset first recovery intensity coefficient; When the driving scenario is urban congestion or low-speed driving and the acceleration j is within the preset acceleration tolerance range, the recovery intensity coefficient λ(k) is set to the preset second recovery intensity coefficient. The multi-source data includes battery SOC. When the battery SOC is lower than a preset battery threshold, the recycling intensity coefficient λ(k) is set to a preset third recycling intensity coefficient.

[0016] Furthermore, based on the motor recovery torque T reg (k) The method for adjusting the kinetic energy recovery intensity is as follows: Obtain the motor torque constant K t According to the motor recovery torque T reg (k) and motor torque constant K t Calculate the motor output current i q,ref (k)Set the maximum recycle current and DC bus current, and then compare the maximum recycle current and DC bus current with the motor output current. i q,ref (k) In contrast, when the motor outputs current i q,ref (k) When the current exceeds the maximum recovery current and the DC bus current, the motor output current... i q,ref (k) It equals the maximum recovery current and the DC bus current; The motor output current i q,ref (k) The calculation expression is as follows ， ; in, i q,ref (k) T is the motor output current. reg (k) represents the motor recovery torque. ω(k) Motor speed Obtain the estimated value of actual motor torque and fault flag bits flag(k) The estimated value of the actual motor torque Fault flag bit flag(k) The battery SOC vector is packaged into a feedback vector y(k), and the desired deceleration a is adjusted based on the feedback vector y(k). des (k) Perform closed-loop correction.

[0017] Furthermore, the method for processing the multi-source data into a system state vector x(k) is as follows: The multi-source data during vehicle operation includes vehicle dynamics data, driver operation data, road and traffic information data, and battery and motor status data; the vehicle dynamics data includes vehicle speed v, acceleration a, and jerk j; the driver operation data includes accelerator pedal opening θ. acc Brake opening θ brk And driver personalized preference parameters; the road and traffic information data includes road gradient γ, distance d from the vehicle in front during driving. f1 and the distance d from the car behind f2 And the relative speed Δv with the vehicle in front; the battery and motor status includes battery SOC, temperature T bat Allowable charging current I ch,max and motor speed ω; The multi-source data is combined into a system state vector, the expression of which is: ; Where x(k) is the system state vector, v(k) is the vehicle speed vector, a(k) is the acceleration vector, j(k) is the jerk vector, and θ acc (k) is the gate opening vector, θ brk (k) is the brake opening vector, γ(k) is the road slope vector, and d f1 (k) is the distance vector between the vehicle and the vehicle in front during the driving process, d f2 (k) is the distance vector to the following vehicle, Δv(k) is the relative velocity vector to the preceding vehicle, and T is the velocity vector to the following vehicle. bat (k) is the temperature vector, I ch,max (k) is the allowable charging current vector, ω(k) is the motor speed vector, and mode(k) is the driver's personalized preference parameter vector.

[0018] Beneficial effects The advantages of this invention are: 1. This invention collects multi-source data during vehicle operation and processes it into a system state vector x(k); it identifies the driving scenario and driving intention based on the system state vector x(k), and simultaneously calculates the desired deceleration a based on the system state vector x(k). des (k) and the target kinetic energy recovery intensity coefficient λ(k); based on the driving scenario, driving intention, and desired deceleration a des (k) and the target kinetic energy recovery intensity coefficient λ(k) output motor recovery torque T reg (k); based on the motor recovery torque T reg (k) Adjust the intensity of kinetic energy recovery; achieve stepless continuous linear adjustment of the intensity of kinetic energy recovery, solve the contradiction between driving comfort and kinetic energy recovery efficiency caused by fixing a small number of kinetic energy recovery intensity gears, and eliminate the feeling of sudden deceleration when releasing the accelerator.

[0019] 2. This invention sets an acceleration tolerance range and a recovery intensity coefficient based on the driving scenario and the acceleration tolerance range, thereby increasing the braking opening range that controls the energy recovery intensity, improving the linear deceleration feel during braking, and enhancing passenger comfort.

[0020] 3. The present invention adjusts the intensity of kinetic energy recovery for different driving scenarios, prioritizing comfort; in low-speed following and hilly / slope scenarios, it ensures both ride comfort and efficient kinetic energy recovery, alleviating driver fatigue from braking; in other scenarios, it ensures both ride comfort and efficient kinetic energy recovery. Attached Figure Description

[0021] Figure 1 This is a flowchart of the adaptive energy management method based on multi-source information fusion according to the present invention; Figure 2This is a flowchart of a specific embodiment of the adaptive energy management method based on multi-source information fusion of the present invention; Figure 3 This is a schematic diagram illustrating the change in kinetic energy recovery intensity in the adaptive energy management method based on multi-source information fusion according to the present invention. Detailed Implementation

[0022] The present invention will be further described below with reference to embodiments, but this does not constitute any limitation on the present invention. Any limited modifications made by any person within the scope of the claims of the present invention are still within the scope of the claims of the present invention.

[0023] See Figure 1-Figure 3 This invention discloses an adaptive energy management method for new energy vehicles based on multi-source information fusion. The management method is as follows: like Figure 1 The diagram shows the collection of multi-source data during vehicle operation via sensors. These sensors include, but are not limited to: speed sensor, gradient sensor, millimeter-wave radar, ignition switch opening sensor, and brake opening sensor; each collecting data in real time. Vehicle dynamics information: vehicle speed data v (unit: km / h), acceleration a (unit: m / s2), jerk j (unit: m / s3).

[0024] Driver operation information: ignition switch opening θ acc (Unit: %), Brake opening θ brk (Unit: %), Driver personalized preference parameters.

[0025] Road and traffic information: Road gradient γ (unit: °), distance d from the vehicle in front during driving. f1 and the distance d from the car behind f2 (Unit: m), relative velocity Δv (unit: km / h).

[0026] Battery and motor status: Battery SOC (%), Temperature T bat (Unit: °C), Allowable charging current I ch,max (Unit: A), Motor speed ω (Unit: rad / min).

[0027] The above information is combined into a system state vector: ; Multi-source data acquisition module M1 The acquired signal is denoised, filtered, and calibrated to ensure input quality. C1 The data is stable and reliable.

[0028] The principle of controlling the intensity of kinetic energy recovery: The intensity of kinetic energy recovery is controlled by controlling the power generation of the motor and thus the resistance torque of the motor, ultimately achieving the purpose of vehicle deceleration and kinetic energy recovery. The formula is as follows: ; in: M Motor resistance torque P Motor power generation ω: Motor speed Electric motor power generation P It is the core of controlling the intensity of kinetic energy recovery. If it is necessary to control the intensity of kinetic energy recovery at a certain moment, the power generation of the motor must be adjusted through the MCU. P , P The larger the value, the higher the kinetic energy recovery intensity, and the more kinetic energy can be recovered.

[0029] The motor speed ω of the vehicle at a certain moment is constant, according to the formula... At a fixed ω, the power generation P Changes will directly affect the drag torque M Power generation capacity P The larger the value, the greater the resistance torque generated by the motor. M The larger the value, the more positively correlated the two are. Among these factors, the motor exhibits a resistance torque. M In the driving experience, this manifests as a dragging sensation when the vehicle speed decreases, or resistance torque. M The larger the value, the more noticeable the "dragging" sensation when the vehicle slows down.

[0030] Electric motor power generation P The kinetic energy recovery intensity is controlled by adjusting circuit parameters via an MCU, and the two extreme states of kinetic energy recovery intensity are also controlled by the power generation. P Control implementation: Kinetic energy recovery intensity is 0: The electronic control system will control the generator power of the motor. P Adjust to 0 and substitute into the formula. have to, M =0, at this time the motor has no resistance torque, the vehicle has no drag and only relies on traditional brakes to decelerate.

[0031] The kinetic energy recovery intensity is max: the electronic control system will control the generator power of the motor. P Pull to the maximum value at the current speed, at this time At this point, the drag torque reaches its maximum, the vehicle decelerates noticeably, and the kinetic energy recovery intensity is at its highest.

[0032] To more accurately explain the sources of discomfort caused to occupants during the kinetic energy recovery process in new energy vehicles, it is necessary to quantitatively analyze typical discomfort mechanisms from the perspectives of vehicle dynamics and ergonomics. The magnitude and rate of change of kinetic energy recovery directly affect the vehicle's longitudinal dynamic response, changes in vehicle posture, and the occupant's vestibular system perception, thus determining ride comfort.

[0033] Occupants are highly sensitive to longitudinal dynamic response. When the intensity of kinetic energy recovery changes abruptly, the vehicle's longitudinal acceleration and jerk will reach peak values. When the jerk is ≥1.5 m / s³, occupants generally experience a noticeable forward lean and discomfort.

[0034] Discontinuous changes in the intensity of the recovery can excite low-frequency longitudinal vibrations in the vehicle at 0.5~5 Hz, of which 1~2 Hz is the most sensitive range for the human body, which can easily cause head shaking, dizziness and fatigue.

[0035] High recovery torque changes increase vehicle pitch rate and pitch acceleration, causing a "nodding" phenomenon and significantly reducing ride comfort.

[0036] If the kinetic energy recovery response is inconsistent with the driver's expected operation (such as the vehicle slowing down significantly even without applying the brakes), it will cause a "mismatch between human and vehicle interaction" and further amplify the discomfort of riding.

[0037] Based on existing research findings, the above examples list typical and significantly uncomfortable working conditions to illustrate the shortcomings of current technology in ensuring passenger comfort and to provide a theoretical basis for the improvement goals of this invention.

[0038] Based on relevant sensor information, combined with preset thresholds and dynamic optimization strategies, dynamic kinetic energy recovery commands of varying intensities are generated for different scenarios. This part is called the data fusion and decision module C1. The core tasks of C1 are: to identify the current driving scenario and driving intention based on the system state vector x(k); and to calculate the expected deceleration a online. des (k) and the target kinetic energy recovery intensity coefficient λ(k), and issue corresponding instructions to implement.

[0039] The core task of C1 is to identify the current driving scenario and driving intention based on the state vector x(k); and to calculate the expected deceleration a online by integrating driver operation, road gradient, and car-following requirements through the expected deceleration calculation model. des By considering factors such as battery status and operating conditions, the trade-off between "comfort" and "energy recovery efficiency" in the current scenario is determined, and an energy recovery priority index β(k) is introduced; the driver's preferred recovery intensity coefficient λ is then used. user By integrating the above dynamic evaluation results, the target kinetic energy recovery intensity coefficient λ(k) is finally obtained.

[0040] The specific functions and implementation methods are as follows: Desired deceleration calculation model This invention calculates the desired deceleration in the following manner: ; in: k1 is the calibration coefficient of the first expected deceleration calculation model, k2 is the calibration coefficient of the second expected deceleration calculation model, k3 is the calibration coefficient of the third expected deceleration calculation model, and k4 is the calibration coefficient of the fourth expected deceleration calculation model. v tar To combine the road slope vector γ(k) and the throttle opening vector θ acc (k), Braking opening vector θ brk (k) Distance vector d between the vehicle in front during driving f1 (k) Distance vector d between the vehicle and the following vehicle during the driving process f2 The target vehicle speed is obtained from (k) and the relative velocity vector Δv(k) with the vehicle in front. This model unifies driver operation, road gradient, and car-following requirements into the desired deceleration index, providing a basis for subsequent recovery intensity calculations.

[0041] Energy recovery priority evaluation This invention introduces an energy recovery priority index coefficient. β(k) This is to determine whether to prioritize comfort or energy recovery in the current scenario. ; in: m SOC (·) This is the normalized membership function that reflects the battery state. m γ (·) The normalized membership function reflects the road slope. m cvcle (·) To reflect the normalized membership function of vehicle operating conditions, m νref (·) To provide a normalized membership function that reflects the driver's individual preferences, ψ SOC As the first weight parameter, ψ γ This is the second weighting parameter. ψ cvcle As the third weighting parameter, ψ νref This is the fourth weighting parameter. β(k)Here, SOC(k) is the energy recovery priority index coefficient, and SOC(k) is the battery SOC vector. v (k) is the vehicle speed vector. a (k) is the acceleration vector, and mode(k) is the driver's personalized preference parameter.

[0042] in, Within this interval β(k) The closer to 1, the more inclined to enhance energy recovery; the closer to 0, the more priority is given to ensuring comfort.

[0043] User base preferences and adaptive recycling intensity fusion To balance the user-perceived sense of "customization" with the system's adaptive capabilities, this invention denotes the driver's set base recovery intensity as λ. user ∈[0%,100%], while the actual recycling intensity λ(k) is dynamically adjusted based on this: ; in, β(k) This is the energy recovery priority index coefficient. a des (k) is the desired deceleration vector. K β The first adjustable gain, K α For the second adjustment gain, β 0 is the first reference point for calibration. α 0 is the second reference point for calibration. ∈[0,1] is the saturation function, and λ(k) is the target kinetic energy recovery intensity coefficient.

[0044] In this way, even if the user only sets a "basic preference", the system will continuously and automatically fine-tune the recycling intensity around it according to the real-time scenario, instead of simply using a "fixed recycling rate".

[0045] The stepless adjustment module C2 for kinetic energy recovery intensity is responsible for receiving the desired deceleration from the data fusion and decision-making module C1. a des (k) and the recovery intensity coefficient λ(k), and convert them into a specific motor recovery torque command T. reg (k) controls the vehicle's kinetic energy recovery process. C2 maximizes passenger comfort while achieving the desired deceleration. Specific functions and implementation methods are as follows: To meet the above requirements, this invention proposes a recovery intensity adjustment strategy that combines model predictive control (MPC) and jerk optimization, as follows: Model Predictive Control (MPC) calculates the optimal energy recovery strategy for future time steps by predicting the vehicle's dynamics model. It optimizes the recovery intensity λ(k) at each time step, minimizing the conflict between acceleration and energy recovery efficiency to ensure a balance between vehicle comfort and energy recovery under different operating conditions. Its objective function is: ; in: J opt To optimize the objective function, a(k+i) For the predicted acceleration, j(k+i) For jerk, ξ1 is the first calibration weighting coefficient, ξ2 is the second calibration weighting coefficient, and ξ3 is the third calibration weighting coefficient. P gen (k+i) The power output of the motor. a des (k+i) is the desired deceleration vector.

[0046] Among them, the objective function is optimized. J opt The method for optimizing the recycling intensity λ(k) at each time step is as follows: a. Based on the vehicle's longitudinal dynamics model, establish the functional relationship between the predicted acceleration a(k+i) and jerk j(k+i) and the recovery intensity coefficient λ(k+i) in the time domain: Predicted acceleration: ; Predicted jerk: ; Predicted power generation: ; Where: △t is the control period; ω(k+i) is the motor speed in the prediction time domain.

[0047] b. At time k in each control cycle, the stepless adjustment module C2 for kinetic energy recovery intensity solves the following finite-time optimization problem: ; Constraints: Recycling strength constraints: ; Acceleration tolerance range constraint: ; Constraint on the rate of change of recycling intensity: ; Battery and motor power constraints:

[0048] C. The above optimization problem is transformed into a standard quadratic programming form and solved using the quadratic programming (QP) algorithm. The specific steps are as follows: Transformation of the standard QP form of the optimization problem The objective function J will be optimized. opt Organized into decision variables The quadratic expression: ; in: It is a symmetric positive definite Hessian matrix, composed of the coefficients of the quadratic terms in the objective function; This is the vector of linear term coefficients; N represents the prediction time domain length.

[0049] The specific structure of the Hessian matrix H is as follows:

[0050] in, , , These are the Jacobian matrices of acceleration, jerk, and power generation versus the recovery intensity coefficient sequence, respectively, derived from the aforementioned prediction model.

[0051] The linear term coefficient vector f is constructed as follows: ; in, To predict the desired deceleration sequence in the time domain.

[0052] Matrix representation of constraints All constraints are uniformly expressed in the form of linear inequality constraints: ; Inequality constraint matrix A ineq sum vector b ineq Includes the following: Upper and lower limits of recycling strength coefficient constraints ; Where IN is an N×N identity matrix, and 1N and 0N are vectors consisting entirely of 1s and 0N, respectively; jerk tolerance range constraint The following is derived from Jmin≤j(k+i)≤Jmax: ; in, j 0 represents the recovery intensity λ(k) at the previous moment. 1) Relevant constant terms; Recycling intensity change rate constraint Depend on Push and get: ; Where D is the difference matrix and e1 is the first unit vector.

[0053] Battery and motor power constraints Depend on The derivation yields:

[0054] QP Solving Algorithm Flow The stepless adjustment module C2 for kinetic energy recovery intensity uses the interior point method or the effective set method to solve the above standard QP problem. The specific solution process is as follows: initialization Set an initial feasible solution Typically, the value is taken as the shifted value of the optimal solution sequence from the previous control cycle; the convergence tolerance ε is set to 10⁻⁶, and the maximum number of iterations K is set to... max =100.

[0055] Constructing the augmented Lagrangian function ; Where μ≥0 are the Lagrange multipliers for inequality constraints, and ν are the Lagrange multipliers for equality constraints.

[0056] Solve the Karush-Kuhn-Tucker (KKT) optimality condition equations: ; The decision variables Λ and Lagrange multipliers (μ,ν) are iteratively updated using Newton's method or a quasi-Newton method until the convergence condition is met: ; When the iteration converges or the maximum number of iterations is reached, the optimal recycling intensity coefficient sequence is output: ; Rolling time-domain execution strategy The rolling time-domain strategy employing model predictive control only takes the first element λ from the optimal sequence. (k) is the output of the actual recovery intensity coefficient at time k, i.e., λ(k) = λ (k).

[0057] At time k+1 in the next control cycle, the new system state feedback x(k+1) is obtained, the prediction model parameters and constraint boundary conditions are updated, and the above QP solution process is repeated. This rolling optimization mechanism ensures that the recovery intensity coefficient λ(k) can respond in real time to changes in vehicle state, driver operation, and road environment, achieving adaptive continuous adjustment.

[0058] Computational efficiency optimization measures To ensure the real-time performance of the QP algorithm on the vehicle controller, the following optimization measures are taken: Hot start strategy: Use the optimal solution of the previous control cycle as the initial guess value of the current cycle to reduce the number of iterations; Sparse matrix utilization: Hessian matrix H and constraint matrix A ineq It has a sparse structure and uses sparse matrix storage and computation to accelerate the solution; Predictive time domain adaptive: The prediction time domain length N is dynamically adjusted according to the vehicle's driving status, with a larger value (N=20~30) for high-speed conditions and a smaller value (N=10~15) for low-speed conditions, balancing the computational burden and control accuracy. Constraint activity prediction: Based on the current operating conditions, predict the constraints that may be active, thereby reducing the number of constraint screenings in the effective set method; The objective function J is optimized using the QP solution method described above. opt It can automatically optimize the recovery intensity coefficient λ at each time step while satisfying all physical and comfort constraints. (k) achieves the best balance between comfort, recovery efficiency and driving intention tracking.

[0059] A rolling time-domain control strategy is adopted, taking only the first element λ in the optimal sequence. (k) is output as the optimal recovery intensity coefficient at time k. At the next time k+1, the new system state feedback is obtained, and the above prediction and optimization process is repeated.

[0060] Through the above optimization mechanism, the objective function J opt The three items in the text each play the following roles: ε1·(aa des ) 2 The λ(k) is guided to ensure that the actual acceleration tracks the desired deceleration, thus guaranteeing the accurate execution of the driving intention; ε2·j 2 Suppress abrupt changes in λ(k). When the accelerometer tends to exceed the tolerance range, automatically reduce the rate of change of λ(k) to eliminate the sense of abrupt change at the source. ε3·P gen 2 Under the premise of satisfying comfort constraints, λ(k) is guided to tend towards a reasonable value to improve energy recovery efficiency.

[0061] The three weighting coefficients ε1, ε2, and ε3 can be dynamically adjusted according to the driving scenario: in the cruising scenario, ε2 is increased to prioritize comfort; in the low-speed following and hilly terrain scenarios, ε1 and ε3 are balanced to take into account both comfort and recycling efficiency; in other scenarios, the values ​​are adaptively adjusted according to the battery SOC status.

[0062] The objective function takes into account the smoothness of the desired deceleration and acceleration, as well as the energy recovery power, so that the system can adaptively adjust the recovery intensity to achieve the predetermined deceleration target, while avoiding abrupt changes in acceleration and improving comfort.

[0063] The formula for jerk is: ; In the recovery strategy, to optimize jerk, we treat jerk as an additional control variable and set a jerk tolerance range. J min , J max If the calculated jerk exceeds this range, the recovery intensity λ(k) will be adjusted to avoid abrupt changes and ensure a smooth transition in acceleration. Increasing the braking opening range for controlling the energy recovery intensity enhances the linear deceleration feel during braking.

[0064] By smoothing the rate of change of the vehicle's longitudinal acceleration, while maintaining the desired deceleration, we can minimize drastic fluctuations in acceleration over a short period of time, thereby reducing discomfort.

[0065] Based on model predictive control (MPC) and jerk optimization, the adjustment of recovery intensity λ(k) is achieved by the following mechanism: When the vehicle is going downhill or traveling at high speed, the recovery intensity λ(k) can be appropriately increased, but the acceleration must be kept within the tolerance range to avoid the vehicle from undergoing excessively rapid longitudinal deceleration. When driving in urban congestion or at low speeds, the recovery intensity needs to be reduced to minimize abrupt deceleration while maintaining smooth acceleration.

[0066] If the battery's SOC is low, the system will automatically reduce the recycling intensity to ensure that the battery is not damaged by excessive charging current.

[0067] The intensity of kinetic energy recovery λ(k) is adjusted in real time through the vehicle's acceleration feedback mechanism. If passengers experience discomfort, the system will adjust the recovery intensity through acceleration optimization to smooth the deceleration process.

[0068] When the vehicle is traveling at low speed, comfort is prioritized, and the recovery intensity λ(k) will be appropriately reduced to avoid sudden braking or abrupt acceleration.

[0069] When the vehicle is going downhill or cruising at high speed, the energy recovery intensity λ(k) will be appropriately increased to improve energy recovery efficiency while still ensuring that the acceleration is within a comfortable range.

[0070] In this invention, "stepless adjustment" is not a simple multi-level interpolation, but is achieved through the following method: The kinetic energy recovery intensity λ(k) is defined as a continuously variable control quantity, which corresponds to the motor recovery torque command. T reg (k) Within the allowable range, any interval can be selected, rather than a preset discrete gear.

[0071] C2 receives the target recovery intensity λ(k) from C1 and uses it as the control target. Through a continuous control algorithm, it adjusts the output in real time to continuously optimize and adjust the recovery intensity λ(k) so that the actual recovery intensity continuously approaches the target value.

[0072] Motor recovery torque command T reg (k) Limiting its rate of change ensures that the recovery intensity changes continuously and smoothly within adjacent control cycles, avoiding sudden changes in recovery torque.

[0073] The formula for transforming λ(k) into Treg(k): ; in: λ(k)∈ [0,1]: Final recovery intensity coefficient after MPC optimization and jerk constraint adjustment T reg,max (k): The maximum allowable recovery torque under the current operating conditions, and: ; T mot,max (k): Motor torque capacity constraint, calculated as follows: ; Where: P gen,max (ω) represents the maximum allowable power output of the motor at the current speed (determined by the motor's external characteristic curve). T bat,max (k): Battery charging capacity constraint

[0074] Among them: U bat (SOC): Current battery voltage; I ch.max (SOC,T bat ): The maximum allowable charging current of the battery; ηgen Generator efficiency; T dec,max (k): Deceleration demand constraint ; Where: m is the total vehicle mass, F roll For rolling resistance, F air For air resistance, F wheel For the wheel radius, I gear For the transmission ratio, η trans For transmission efficiency.

[0075] In cruising scenarios, comfort is prioritized; in low-speed following and hilly terrain scenarios, ride comfort and efficient energy recovery are ensured, reducing driver fatigue from braking; in other scenarios, ride comfort and efficient energy recovery are maintained.

[0076] This allows for stepless, continuous, and linear adjustment of the kinetic energy recovery intensity, resolving the conflict between driving comfort and kinetic energy recovery efficiency caused by fixing a small number of kinetic energy recovery intensity levels, and eliminating the feeling of sudden deceleration when releasing the accelerator.

[0077] Execution and Feedback Module M2 Module for stepless adjustment of kinetic energy recovery intensity C2 Output T reg (k) Physical execution and closed-loop feedback are the key execution layer structures for realizing the stepless adjustment of kinetic energy recovery, adaptive control, and ride comfort of this invention. Specific functions and implementation methods are as follows: The motor control unit (MCU) primarily executes the regenerative braking torque command, which is to... C2 Output target motor recovery torque T reg (k) is the core of converting motor current commands and implementing closed-loop control.

[0078] Kinetic energy recovery intensity infinitely adjustable module C2 The target recovery torque at time K is calculated to be T. reg (k), the MCU is based on the motor torque constant K t And parameters such as efficiency, to determine the motor output current. i q,ref (k) : ; i q,ref (k) T is the motor output current. reg (k) represents the motor recovery torque, and ω(k) represents the motor speed.

[0079] Simultaneously combining the capabilities of the motor and inverter, i q,ref(k) Limit the current to ensure it does not exceed the maximum permissible recovery current and DC bus current: ; The MCU internally employs a fast current PI / vector control loop for high-frequency closed-loop control of the current (10 kHz). To complement the jerk-based comfort control in the stepless adjustment module C2 for kinetic energy recovery intensity, the MCU further incorporates current change rate limiting and torque slope limiters at the execution layer to ensure the actual recovered torque T of the motor. reg The change in (k) is continuous and smooth: ; in, ΔT reg,max This is the difference between the actual recovered torque limit of the motor and the actual torque limit.

[0080] With the above restrictions, even if the command from the higher level changes rapidly, the actual torque will still transition smoothly according to the set slope, further suppressing the peak of longitudinal acceleration and improving ride comfort.

[0081] Feedback Measurement and Condition Monitoring It provides key feedback signals for the data fusion and decision-making module C1 and the stepless adjustment module for kinetic energy recovery intensity C2, forming the basis for achieving closed-loop adaptive control. The main quantities it collects and monitors include, but are not limited to: Actual motor torque estimate (Estimated using current and motor models) Key states of the battery, including voltage, current, state of charge (SOC), and temperature. Fault flags and diagnostic results (sensor failure, motor overheating, battery overcharge, etc.) The above information is packaged into a feedback vector y(k) in real time: .

[0082] And periodically send it to C2 via the vehicle network for: The data fusion and decision-making module C1 and the stepless adjustment module C2 for kinetic energy recovery intensity perform closed-loop correction of the desired deceleration / acceleration.

[0083] a. Definition and composition of feedback vector: ; in: : This is the estimated value of the actual motor torque at time k, obtained from the motor current and the motor model. : is the battery state of charge vector at time k; flag(k): is the fault flag vector at time k, including diagnostic results such as sensor failure, motor overheating, and battery overcharging.

[0084] b. The overall framework of closed-loop correction: This invention employs a multi-level feedback correction mechanism, using each component of the feedback vector y(k) to apply to the desired deceleration a. des Different correction elements (k) form a complete closed-loop control system. The formula for calculating the corrected desired deceleration is: ; Where: a des (k) represents the expected deceleration initially calculated by the data fusion and decision-making module C1; η torque (k) is the torque tracking correction coefficient; η is the battery state correction factor. flag (k) is the fail-safe correction factor.

[0085] c. Torque tracking correction factor η torque Calculation of (k): Torque tracking correction coefficient η torque (k) is used to compensate for the deviation between the actual output torque of the motor and the theoretical required torque. Its calculation expression is as follows: ; in: This is to normalize the torque tracking error; The gain is used for torque ratio correction, with a value range of [0.1, 0.5]. The gain is the torque integral correction gain, with a value range of [0.01, 0.1]; Δt is the control period, in seconds.

[0086] To ensure the stability of the correction coefficient, for η torque (k) Perform amplitude limiting: ; in: =0.7, =1,sat(·) is a saturated function.

[0087] d. Battery state correction factor η SOC Calculation of (k): Battery state correction factor η SOC (k) is used to dynamically adjust the desired deceleration based on the battery's state of charge to avoid overcharging or undercharging. Its calculation expression is: ; Among them: SOC opt,min This represents the lower limit of the battery's optimal operating range, with a preferred value of 20%; SOC opt,max This represents the upper limit of the battery's optimal operating range, with a preferred value of 80%. The gain is adjusted for high charge levels, with a value range of [0.3, 0.6]. The gain is adjusted for low battery levels, and its value ranges from [0.5, 0.8].

[0088] When the battery SOC is too high, the expected deceleration is reduced to decrease energy recovery and prevent overcharging; when the battery SOC is too low, the expected deceleration is reduced to protect the battery and avoid damage from high-current charging.

[0089] e. Fail-safety correction factor η flag Calculation of (k): Fail-safety correction factor η flag (k) is used to quickly reduce the desired deceleration when a system fault is detected, ensuring driving safety. Its calculation expression is: ; Among them: flag i (k) is the flag bit for the i-th type of fault, taking a value of 0 or 1, where 0 indicates no fault and 1 indicates a fault; α i The corrected weight for the i-th type of fault is [0.3, 1.0], with a value range of [0.3, 1.0]; N is the total number of fault types; The fault types include, but are not limited to: motor overheating fault: α temp =0.5; Battery overcharge fault: α bar =0.5; Sensor failure fault: α sensor =0.5; Communication failure: α comm =0.5.

[0090] When a serious fault is detected, η flag (k) can be reduced to 0, at which point the desired deceleration is... The system exits the kinetic energy recovery mode and switches to traditional mechanical braking.

[0091] The beneficial effects of closed-loop correction: The above closed-loop correction mechanism achieves the following beneficial effects: Improved torque tracking accuracy: achieved through torque tracking correction coefficient η torque (k) Real-time compensation for the deviation between the actual output of the motor and the theoretical requirement, so that the actual deceleration can more accurately track the expected deceleration, and the tracking error is reduced by more than 30%; Enhanced battery protection: achieved through a battery state correction coefficient η SOC (k) Dynamically adjust the energy recovery intensity according to the battery SOC to avoid overcharging or undercharging the battery and extend the battery life; Improved safety and reliability: through the fail-safe correction factor η flag (k) quickly reduces or exits kinetic energy recovery when a system fault is detected, ensuring driving safety.

[0092] When the ride comfort index is detected to deviate from the set range, the adaptive adjustment of the higher-level control strategy is triggered. The execution and feedback module M2 is responsible for "implementing the strategy". With the synergy of MCU, feedback measurement and status monitoring, it accurately executes and corrects these target quantities in a closed loop. The execution and feedback module M2 not only ensures the physical feasibility of stepless adjustment of kinetic energy recovery intensity, but also together with the adaptive control of C1 and C2, it forms a complete closed-loop system oriented towards ride comfort.

[0093] An adaptive kinetic energy recovery system based on multi-source information fusion includes: Multi-source data acquisition module M1 It is used to acquire multi-source data and output multi-source data signals.

[0094] Data Fusion and Decision Module C1 This module is used to receive multi-source data signals and apply the aforementioned adaptive kinetic energy recovery method based on multi-source information fusion to control the stepless adjustment module of kinetic energy recovery intensity according to the multi-source data. C2 and execution and feedback module M2 To achieve stepless, continuous, and linear adjustment of the kinetic energy recovery intensity, the contradiction between driving comfort and kinetic energy recovery efficiency caused by fixing a small number of kinetic energy recovery intensity levels is resolved, and there is no sudden deceleration when releasing the accelerator.

[0095] Kinetic energy recovery intensity infinitely adjustable module C2， Used to convert policy quantities into continuous recycling instructions.

[0096] Execution and Feedback Module M2， Execute and provide feedback on the actual operating status.

[0097] With fewer modules, data can be lightweighted and costs reduced; the modules communicate and connect sequentially to form a closed-loop adaptive kinetic energy recovery system. M1 Responsible for collecting multi-source information, data fusion and decision-making modules C1 The module responsible for scene recognition and recovery strategy decision-making, and infinitely adjustable kinetic energy recovery intensity C2 The policy quantity is converted into continuous recycling instructions, and the execution and feedback module is used. M2 Execute and provide feedback on the actual operating status.

[0098] like Figure 2 As shown, the following provides five examples of adaptive energy management in different scenarios to facilitate understanding: Example 1 (Cruising): The speed sensor detects the current vehicle speed v1∈[50,+∞) and maintains it for 5s. The millimeter-wave radar detects the distance between the vehicle and the vehicle in front. d f1 ∈[80,+∞), C1Determine that the vehicle is in a cruising scenario and make the following decisions: Switch opening θ acc With brake opening θ brk When both are 0, the data fusion and decision-making module C1 Issue "a" des1 The instruction "(k)=0, λ1(k)=0" indicates the stepless adjustment module for kinetic energy recovery intensity. C2 The response command outputs "T" reg1 (k)=0”, Execution and Feedback Module M2 Responding to the command yields " i q,ref1 (k) The system displays "=0" and provides continuous feedback, ensuring the vehicle experiences no dragging sensation due to regenerative braking, and its driving state is similar to that of a gasoline-powered vehicle coasting. This feature guarantees driving comfort.

[0099] Example 2 (Low-speed following): The speed sensor detects the current vehicle speed v2∈[0,50) and maintains it for 5s. The millimeter-wave radar detects the distances between the vehicle and the vehicle in front and the vehicle behind, which meet the following conditions. d f1 ∈[5,80)∪ d f2 ∈[5,80), Data Fusion and Decision Module C1 Determine that the vehicle is in a low-speed following scenario and make the following decisions: Switch opening θ acc With brake opening θ brk When both are 0, the data fusion and decision-making module C1 Issue "a" des2 The instruction "λ(k) gradually increases" is used to adjust the kinetic energy recovery intensity steplessly. C2 The response command outputs "T" reg2 (k) gradually increases adaptively from 0 to Max”, execution and feedback module M2 Responding to the command yields " i q,ref2 (k) It gradually increases in size and provides continuous feedback. This feature ensures driving comfort and fully recovers kinetic energy. Dynamically monitor the distance between the vehicle and the vehicle in front during kinetic energy recovery. d f1 Distance to the car behind d f2 When the distance between the vehicle and the vehicle in front d f1 Distance between the vehicle and the vehicle behind d f2 Changes occur during this process, affecting the data fusion and decision-making modules. C1 Issue "desired deceleration target value a"des3 The instruction "(k), kinetic energy recovery intensity target value λ3(k)" is used by the stepless adjustment module for kinetic energy recovery intensity. In response to the command, the vehicle's kinetic energy recovery intensity adaptively changes from the current value to the target T. reg3 (k)”, Execution and Feedback Module M2 The response command resulted in "adaptive change from the current current to the target current". i q,ref3 (k) "And continuously provide feedback to maintain a safe distance from the vehicle in front." d f1 Distance to the car behind d f2 The variation range is within ±5, dynamically controlling the intensity of kinetic energy recovery and adaptively controlling the distance between vehicles. This function reduces driver fatigue from braking, ensures driving comfort, and adaptively and dynamically adjusts the intensity of kinetic energy recovery to fully recover kinetic energy.

[0100] Example 3 (Hillary Slope): When the slope sensor detects a road slope γ∈[-∞,-10)∪[10,+∞) and maintains it for five seconds, the data fusion and decision module... C1 Determine that the vehicle is in a hilly terrain scenario and make the following decisions: Switch opening θ acc With brake opening θ brk When all values ​​are 0 and the road slope γ∈ [10, +∞), the data fusion and decision module... C1 Issue "a" des4 The instruction "(k)=0, λ4(k)=0" indicates the stepless adjustment module for kinetic energy recovery intensity. The response command outputs "T" reg4 (k)=0”, Execution and Feedback Module Responding to the command yields " i q,ref3 (k)=0 "And continuous feedback shows that the vehicle does not experience any dragging sensation due to kinetic energy recovery, and its driving state is close to that of a gasoline-powered vehicle coasting. This function ensures driving comfort."

[0101] Switch opening θ acc With brake opening θ brk All values ​​are 0 and the road slope γ∈(-∞,-10], data fusion and decision module C1 Issue "desired deceleration target value a" des5 The instruction "(k), kinetic energy recovery intensity target value λ5(k)" is used by the stepless adjustment module for kinetic energy recovery intensity. C2 In response to the command, the vehicle's kinetic energy recovery intensity adaptively changes from the current value to the target T. reg5 (k)” ，Execution and Feedback Module M2 The response command resulted in "adaptive change from the current current to the target current". i q,ref5 (k) "And it provides continuous feedback. By dynamically controlling the intensity of kinetic energy recovery, the adaptive control v3 varies within a range of ±2, dynamically adjusting the intensity of kinetic energy recovery to achieve the purpose of constant speed descent on downhill slopes. This function reduces driver fatigue, ensures driving comfort, and fully recovers kinetic energy."

[0102] Example 4 (Other): When none of the sensor data meets the pre-set threshold conditions, the data fusion and decision-making module... C1 Determine if the vehicle is in another scenario and make the following decisions: Switch opening θ acc With brake opening θ brk When both are 0, the data fusion and decision-making module C1 Issue "desired deceleration target value a" des6 The instruction "(k), kinetic energy recovery intensity target value λ6(k)" is used by the stepless adjustment module for kinetic energy recovery intensity. C2 In response to the command, the vehicle's kinetic energy recovery intensity adaptively changes from the current value to the target T. reg6 (k)”, Execution and Feedback Module M2 The response command resulted in "adaptive change from the current current to the target current". i q,ref6 (k) "And provides continuous feedback. This feature ensures driving comfort and fully recovers kinetic energy."

[0103] Example 5 (Custom): The system provides basic strength λ user A fixed value of kinetic energy recovery intensity between [0%, 100%] (accuracy 0.1%) is available for selection; data fusion and decision module. C1 Issue "desired deceleration target value a" des7 The instruction "(k), kinetic energy recovery intensity target value λ7(k)" is used by the stepless adjustment module for kinetic energy recovery intensity. C2 In response to the command, the vehicle's kinetic energy recovery intensity adaptively changes from the current value to the target T. reg7 (k)”, Execution and Feedback Module M2 The response command resulted in "adaptive change from the current current to the target current". i q,ref7 (k) "And provide continuous feedback. Users can set a fixed kinetic energy recovery intensity according to their personal preferences, adapting to different usage scenarios and driving interests."

[0104] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several modifications and improvements without departing from the structure of the present invention, and these will not affect the effectiveness of the implementation of the present invention or the practicality of the patent.

Claims

1. A new energy vehicle adaptive energy management method based on multi-source information fusion, characterized in that, The method includes collecting multi-source data during vehicle operation, processing the multi-source data into a system state vector x(k); identifying the driving scenario and driving intention based on the system state vector x(k), and simultaneously calculating the desired deceleration a based on the system state vector x(k). des (k) and the target kinetic energy recovery intensity coefficient λ(k); based on the driving scenario, driving intention, and desired deceleration a des (k) and the target kinetic energy recovery intensity coefficient λ(k) output motor recovery torque T reg (k); based on the motor recovery torque T reg (k) Adjust the intensity of kinetic energy recovery.

2. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 1, characterized in that, The desired deceleration a is calculated based on the system state vector x(k). des The method for (k) is that the system state vector x(k) includes the road slope vector γ(k) and the throttle opening vector θ. acc (k), Braking opening vector θ brk (k) Distance vector d between the vehicle in front and the vehicle during driving f1 (k) Distance vector d between the vehicle and the following vehicle during the driving process f2 (k) and the relative velocity vector Δv(k) with the vehicle in front, based on the road slope vector γ(k) and the throttle opening vector θ acc (k), Braking opening vector θ brk (k) Distance vector d between the vehicle in front and the vehicle during driving f1 (k) Distance vector d between the vehicle and the following vehicle during the driving process f2 The desired deceleration a is calculated from (k) and the relative velocity vector Δv(k) with the vehicle in front. des (k); The method for calculating the target kinetic energy recovery intensity coefficient λ(k) based on the system state vector x(k) is as follows: the system state vector x(k) further includes the battery SOC vector SOC(k), road slope vector γ(k), vehicle speed vector v(k), acceleration vector a(k), and driver personalized preference parameter mode(k). The energy recovery priority index coefficient is calculated based on the battery SOC vector SOC(k), road slope vector γ(k), vehicle speed vector v(k), acceleration vector a(k), and driver personalized preference parameter mode(k). β(k) Set the foundation strength λ user (k), based on the aforementioned basic strength λ user (k) Energy recovery priority index coefficient β(k) and expected deceleration a des The target kinetic energy recovery intensity coefficient λ(k) is calculated.

3. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 2, characterized in that, The desired deceleration a des The expression for calculating (k) is, ； Where k1 is the calibration coefficient of the first expected deceleration calculation model, k2 is the calibration coefficient of the second expected deceleration calculation model, k3 is the calibration coefficient of the third expected deceleration calculation model, and k4 is the calibration coefficient of the fourth expected deceleration calculation model. a des (k) is the desired deceleration vector. γ (k) is the road slope vector. v (k) is the vehicle speed vector. θ acc (k) is the gate opening vector, θ brk (k) is the braking opening vector. v tar (k) To combine the road slope vector γ(k) and the throttle opening vector θ acc (k), Braking opening vector θ brk (k) Distance vector d between the vehicle in front and the vehicle during driving f1 (k) Distance vector d between the vehicle and the following vehicle during the driving process f2 The target vehicle speed is obtained from (k) and the relative velocity vector Δv(k) with the vehicle in front.

4. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 3, characterized in that, The energy recovery priority index coefficient β(k) The calculation expression is, ； in, μ SOC (·) This is the normalized membership function that reflects the battery state. μ γ (·) The normalized membership function reflects the road slope. μ cvcle (·) To reflect the normalized membership function of vehicle operating conditions, μ νref (·) To provide a normalized membership function that reflects the driver's individual preferences, ψ SOC As the first weight parameter, ψ γ This is the second weighting parameter. ψ cvcle As the third weighting parameter, ψ νref This is the fourth weighting parameter. β(k) Here, SOC(k) is the energy recovery priority index coefficient, and SOC(k) is the battery SOC vector. v (k) is the vehicle speed vector. a (k) is the acceleration vector, and mode(k) is the driver's personalized preference parameter.

5. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 4, characterized in that, The calculation expression for the target kinetic energy recovery intensity coefficient λ(k) is as follows: ； in, β(k) This is the energy recovery priority index coefficient. a des (k) is the desired deceleration vector. K β The first adjustable gain, K α For the second adjustment gain, β 0 is the first reference point for calibration. α 0 is the second reference point for calibration. λ(k) is the saturation function, and λ(k) is the target kinetic energy recovery intensity coefficient.

6. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 1, characterized in that, Based on the driving scenario, driving intention, and desired deceleration a des (k) and the target kinetic energy recovery intensity coefficient λ(k) output motor recovery torque command T reg The method for (k) is as follows: Based on the jerk j(k) and acceleration a(k), based on the desired deceleration a des (k) The optimization objective function is calculated. J opt Through the optimization objective function J opt The recovery intensity coefficient λ(k) is initially optimized, and then further adjusted based on the driving scenario, driving intention, and jerk j(k) to generate the motor recovery torque command T. reg (k).

7. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 6, characterized in that, The optimization objective function J opt The calculation expression is, ； in, J opt To optimize the objective function, a(k+i) For the predicted acceleration, j(k+i) For jerk, ξ1 is the first calibration weighting coefficient, ξ2 is the second calibration weighting coefficient, and ξ3 is the third calibration weighting coefficient. P gen (k+i) The power output of the motor. a des (k+i) is the desired deceleration vector, and N is the prediction time domain length.

8. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 7, characterized in that, Based on the driving scenario, driving intention, and acceleration j(k), the initially optimized recovery intensity coefficient λ(k) is further adjusted to form the motor recovery torque T. reg The method for (k) is as follows: When the driving scenario is a vehicle going downhill or traveling at high speed and the acceleration j is within the preset acceleration tolerance range, the recovery intensity coefficient λ(k) is set to the preset first recovery intensity coefficient; When the driving scenario is urban congestion or low-speed driving and the acceleration j is within the preset acceleration tolerance range, the recovery intensity coefficient λ(k) is set to the preset second recovery intensity coefficient. The multi-source data includes battery SOC. When the battery SOC is lower than a preset battery threshold, the recycling intensity coefficient λ(k) is set to a preset third recycling intensity coefficient.

9. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 1, characterized in that, According to the motor recovery torque T reg (k) The method for adjusting the kinetic energy recovery intensity is as follows: Obtain the motor torque constant K t According to the motor recovery torque T reg (k) and motor torque constant K t Calculate the motor output current i q,ref (k) ; Set the maximum recycle current and DC bus current, and then compare the maximum recycle current and DC bus current with the motor output current. i q,ref (k) In contrast, when the motor outputs current i q,ref (k) When the current exceeds the maximum recovery current and the DC bus current, the motor output current... i q,ref (k) It equals the maximum recovery current and the DC bus current; The motor output current i q,ref (k) The calculation expression is, ； in, i q,ref (k) T is the motor output current. reg (k) represents the motor recovery torque, and ω(k) represents the motor speed. Obtain the estimated value of actual motor torque and fault flag bits flag(k) The estimated value of the actual motor torque Fault flag bit flag(k) The battery SOC vector is packaged into a feedback vector y(k), and the desired deceleration a is adjusted based on the feedback vector y(k). des (k) Perform closed-loop correction.

10. The adaptive energy management method for new energy vehicles based on multi-source information fusion according to claim 1, characterized in that, The method for processing the multi-source data into a system state vector x(k) is as follows: The multi-source data during vehicle operation includes vehicle dynamics data, driver operation data, road and traffic information data, and battery and motor status data; the vehicle dynamics data includes vehicle speed v, acceleration a, and jerk j; the driver operation data includes accelerator pedal opening θ. acc Brake opening θ brk And driver personalized preference parameters; the road and traffic information data includes road gradient γ, distance d from the vehicle in front during driving. f1 and the distance d from the car behind f2 And the relative speed Δv with the vehicle in front; the battery and motor status includes battery SOC, temperature T bat Allowable charging current I ch,max and motor speed ω; The multi-source data is combined into a system state vector, the expression of which is: ； Where x(k) is the system state vector, v(k) is the vehicle speed vector, a(k) is the acceleration vector, j(k) is the jerk vector, and θ acc (k) is the gate opening vector, θ brk (k) is the brake opening vector, γ(k) is the road slope vector, and d f1 (k) is the distance vector between the vehicle and the vehicle in front during the driving process, d f2 (k) is the distance vector to the following vehicle, Δv(k) is the relative velocity vector to the preceding vehicle, and T is the velocity vector to the following vehicle. bat (k) is the temperature vector, I ch,max (k) is the allowable charging current vector, ω(k) is the motor speed vector, and mode(k) is the driver's personalized preference parameter vector.

Images