Electric vehicle parking motor torque adjustment method

Abstract

This application relates to the field of electric vehicle technology, and in particular to a method for regulating the torque of a parking motor in electric vehicles. It aims to address the problems of insufficient parking force control accuracy, lag response, and low energy consumption efficiency in existing electric vehicle parking systems under complex operating conditions. This method establishes and updates a vehicle-environment coupled dynamic model in real time by collecting multi-dimensional vehicle operating status, driver commands, and environmental perception data. Then, using an adaptive model predictive control (AMPC) strategy, based on the dynamic model and target parking requirements, it predicts the future movement trend of the vehicle, optimizes the calculation of the optimal torque command for the parking motor within the current control cycle, and drives the parking motor to generate the corresponding torque. This application achieves high precision, high adaptability, high safety, and high stability in parking torque regulation, effectively improving energy efficiency and component lifespan, and enhancing the level of intelligence.

Description

Technical Field

[0001] This application relates to the field of electric vehicle technology, and in particular to a method for adjusting the torque of a parking motor in an electric vehicle. Background Technology

[0002] The electric vehicle parking motor torque regulation method refers to achieving stable parking and safe starting of the vehicle under various operating conditions by precisely controlling the output torque of the parking motor. This method aims to solve problems such as vehicle slippage, impact, insufficient parking force, or overload that may occur during parking. It proposes an optimized torque regulation strategy. This method designs a collaborative mechanism between the parking motor and the vehicle control system, allowing the parking torque to be flexibly adjusted and applied according to road conditions, gradient, load, and user needs. In this method, the parking motor receives data from vehicle status sensors, calculates the required parking force, and precisely outputs the corresponding torque to achieve reliable braking and fixation of the vehicle. Each parking motor, based on its type and characteristics, independently or collaboratively performs torque output tasks, thereby improving the safety and intelligence level of the overall parking system. This method also optimizes the comfort and response efficiency of parking braking by adjusting the torque output curve and release timing.

[0003] Existing technologies for parking torque control largely rely on preset parameters, lacking the ability to adapt to dynamic changes in actual vehicle load and road friction coefficient. This makes it difficult to effectively ensure parking safety on extreme slopes or slippery surfaces. For example, when a heavily loaded vehicle is parked on a steep slope, the preset torque may not be sufficient to completely overcome gravity, causing the vehicle to slide. However, due to the lack of a real-time load assessment mechanism, the risk of vehicle slippage cannot be predicted. Parking motor torque output is mostly based on a fixed pulse width modulation (PWM) period rather than the actual temperature evolution of the motor, making it difficult to accurately reflect the motor's thermal state and output limits, thus hindering motor lifespan and continuous parking capability. The lack of a coordination mechanism between parking commands and the vehicle control module can lead to conflicts between the parking motor and the main drive motor in emergency parking or frequent start-stop scenarios, resulting in energy loss under high concurrency. The lack of a comparison mechanism between parking status logs and environmental parameter data makes it difficult to detect suboptimal parking strategies, resulting in unnecessary energy waste. In terms of maintaining parking force, existing technologies mainly rely on constant torque output, ignoring the subtle fluctuations in parking force under different environments and energy-saving optimizations. This makes it difficult to adapt to the control needs under dynamic changes. The problem is particularly prominent in multi-condition parking of electric vehicles, directly affecting overall parking safety and user experience. Summary of the Invention

[0004] This invention relates to the field of electric vehicle technology, and in particular discloses a method for adjusting the torque of an electric vehicle parking motor, aiming to solve the problems of insufficient parking force control precision, lag response, and low energy consumption efficiency in existing electric vehicle parking systems under complex operating conditions. Traditional electric vehicle parking systems, whether through mechanical handbrake linkage or simple electronic parking brake systems, often employ preset fixed torque or simple feedback control strategies based on basic sensors. This method struggles to achieve precise torque control of the parking motor in complex environments such as hill starts, changes in road surface adhesion coefficient, or dynamic changes in vehicle load. This can lead to excessive parking force causing energy waste and component wear, or insufficient parking force causing unexpected vehicle slippage, thus affecting parking safety and user experience. Furthermore, existing systems typically lack deep perception and predictive capabilities regarding vehicle and environmental conditions, and cannot dynamically adjust torque output to adapt to rapidly changing actual operating conditions, resulting in an unsmooth and inefficient parking process.

[0005] To address the aforementioned technical problems, this invention proposes a method for regulating the parking motor torque of an electric vehicle based on multi-source information fusion and adaptive predictive control. This invention constructs a dynamic vehicle-environment coupling model by real-time acquisition of multi-dimensional vehicle state and environmental parameters. It then utilizes an adaptive model predictive control (AMPC) strategy to accurately predict the vehicle's future motion trends and optimize the calculation of the optimal torque output sequence of the parking motor under specific parking requirements. This significantly improves the accuracy, response speed, and energy efficiency of parking torque regulation while ensuring parking safety and stability. This invention not only considers the vehicle's current state but also incorporates predictions of future operating conditions, achieving proactive parking torque management.

[0006] According to one aspect of the present invention, a method for adjusting the torque of a parking motor in an electric vehicle is provided, the method comprising:

[0007] The system collects real-time vehicle operating status data, driver operation commands, and environmental perception data. The vehicle operating status data includes, but is not limited to, vehicle speed, acceleration, slope angle, battery level, and drive motor torque. The driver operation commands include, but are not limited to, parking request, parking release request, and parking mode selection. The environmental perception data includes, but is not limited to, the estimated value of the road surface adhesion coefficient.

[0008] Based on the real-time collected data, a current vehicle-environment coupled dynamic model is established, and the parameters of the model are updated in real time. The vehicle-environment coupled dynamic model is used to describe the parking mechanical characteristics of the vehicle under different slope, road surface adhesion coefficient, and load conditions.

[0009] Based on the driver's operating instructions and the vehicle-environment coupled dynamic model, the target parking force requirement is determined. This target parking force requirement reflects the parking force required to ensure the vehicle remains stationary or moves smoothly.

[0010] An adaptive model predictive control strategy is employed, based on the vehicle-environment coupled dynamic model and the target parking force requirement, to predict the future state of the vehicle within a preset prediction time domain, and to optimize the calculation of the optimal torque command for the parking motor in the current control cycle. This optimal torque command aims to make the vehicle state infinitely close to the target parking state while minimizing energy consumption and torque fluctuations.

[0011] The optimal torque command is sent to the parking motor controller to drive the parking motor to generate the corresponding torque to enable the vehicle to be parked or released from parking.

[0012] Furthermore, during the real-time data acquisition phase, this invention acquires the vehicle's real-time attitude and kinematic information through in-vehicle sensors (such as inertial measurement units, wheel speed sensors, and brake pressure sensors). Simultaneously, it combines this data with the vehicle's CAN bus data to obtain battery state of charge (SOC) and energy consumption data from the battery management system, as well as drive torque information from the drive motor controller. For environmental perception data, this invention can utilize sensor data from onboard cameras, millimeter-wave radar, and other sensors, combined with high-precision map information, to accurately estimate the slope angle where the vehicle is located using visual recognition algorithms or radar ranging and speed measurement algorithms. The road surface adhesion coefficient can be estimated through real-time correlation analysis of tire slip ratio and drive / braking torque, or through inverse estimation using a vehicle dynamics model.

[0013] In the stage of establishing a vehicle-environment coupled dynamic model and updating parameters in real time, this invention constructs a comprehensive model capable of describing the vehicle's longitudinal motion, lateral sideslip trend, and the transmission characteristics of the parking mechanism. This model not only considers fixed parameters such as vehicle mass, wheelbase, and center of gravity height, but more importantly, it can dynamically absorb and update key parameters such as actual vehicle mass (considering load changes), road rolling resistance coefficient, air resistance coefficient, and the actual transmission efficiency of the parking mechanism. Real-time parameter updates can be achieved through state estimation methods such as Kalman filtering, extended Kalman filtering, or unscented Kalman filtering. Iterative corrections are made based on the difference between the vehicle's actual motion response and the model's predicted output to ensure the model's accuracy and timeliness.

[0014] The target parking force requirement is determined based on a comprehensive assessment of the driver's intentions and the current vehicle-environment state. When the driver requests to park, the system calculates the minimum parking force required to overcome the gravitational component and road rolling resistance, based on the current slope of the vehicle and the estimated road adhesion coefficient. To improve safety, a safety margin is usually added to the minimum parking force. When the driver releases the parking brake, the target parking force gradually decreases to zero, or maintains a low level of auxiliary braking force under specific conditions to ensure a smooth and shock-free vehicle start.

[0015] The core of this invention is the adaptive model predictive control strategy. Its basic idea is to predict the future motion state of the vehicle within a finite prediction time domain using an updated vehicle-environment coupled dynamic model in each control cycle. Based on this, an optimization algorithm solves a quadratic programming problem to obtain the optimal torque sequence of the parking motor within the future control time domain. This optimization problem aims to minimize the deviation between the parking state and the target state, while considering the stability of the control quantity (i.e., minimizing the rate of torque change) and minimizing energy consumption, and must strictly meet parking safety-related constraints, such as maximum parking force, minimum release force, and maximum permissible slip during parking. The adaptive aspect is reflected in the fact that the model parameters are corrected in each control cycle, ensuring that the controller always predicts and optimizes based on a model that most closely approximates the actual operating conditions.

[0016] To clearly illustrate the dynamic characteristics of a vehicle when parked, this invention uses the following simplified longitudinal dynamics model of the vehicle as the basis for model predictive control:

[0017]

[0018] in,

[0019] This represents the total mass of the electric vehicle and is a variable parameter that is affected by the number of occupants and the amount of cargo carried. This indicates the vehicle's longitudinal acceleration; This indicates the equivalent parking force applied to the wheels by the parking mechanism; Represents the gravitational acceleration constant; This indicates the current slope angle of the road surface; uphill is positive and downhill is negative. Indicates the rolling resistance of the road surface; This represents air resistance. This formula describes the longitudinal motion of a vehicle under parking force after overcoming the gravity component of the slope, rolling resistance, and air resistance. Under stable parking conditions, Approaching zero.

[0020] Furthermore, the adaptive model predictive control (AMPC) strategy determines the optimal parking motor torque sequence by solving an online optimization problem. The objective function of this optimization problem can be expressed as:

[0021]

[0022] in,

[0023] In the current control cycle The cost function that minimizes internal demand; It is the length of the predicted time domain; It controls the length of the time domain; Indicates at time Predicted future moments Vehicle status output (e.g., vehicle speed or parking force); This corresponds to the target reference state; It is a weighted matrix of state errors, used to measure the penalty for deviation between the predicted state and the target state; Indicates future time The change in the parking motor torque command, i.e. ; It is a weighted matrix of the rate of change of the control variable, used to penalize the intensity of the control action in order to ensure stability; This indicates the estimated energy consumption of the parking motor during this control cycle; These are the weighting coefficients for energy consumption. This objective function comprehensively considers parking accuracy, stability, and energy efficiency, and adjusts the weight matrix accordingly. and and coefficients Trade-offs can be made among these performance metrics.

[0024] In practical applications, there is a transmission ratio relationship between the torque output by the parking motor and the parking force of the vehicle. To achieve precise parking force control, it is necessary to calculate the torque required by the parking motor in reverse, based on the parking force requirement obtained from the optimization problem described above. This relationship can be approximately expressed as:

[0025]

[0026] in, This refers to the torque output by the parking motor; The parking force to be applied to the wheels; The effective radius of the wheel; The reduction ratio of the parking mechanism; The transmission efficiency of the parking mechanism is considered. In actual operation, the reduction ratio and transmission efficiency of the parking mechanism may fluctuate slightly due to wear or temperature changes. This invention ensures the accuracy of torque conversion by including real-time estimation and updating of these parameters in the vehicle-environment coupled dynamic model.

[0027] Compared with existing technologies, the electric vehicle parking motor torque adjustment method provided by this invention, by introducing multi-source information fusion, a dynamic vehicle-environment coupling model, and an adaptive model predictive control strategy, has the following beneficial effects:

[0028] High precision and high adaptability: This invention can sense vehicle status and environmental changes in real time and dynamically update vehicle model parameters, so that the parking motor torque adjustment can accurately match the actual working conditions. It exhibits excellent adaptability and high-precision parking capability, especially in complex environments such as slopes, different road surface adhesion coefficients and load changes.

[0029] Enhanced safety: By predicting the future movement trend of the vehicle and performing proactive control, it effectively avoids accidental slippage caused by insufficient parking force or impact caused by excessive parking force, significantly improving the safety and reliability of the parking process.

[0030] Stability and comfort optimization: The adaptive model predictive control strategy can optimize the rate of change of control variables, making the application and release of parking torque smoother, reducing vehicle vibration and impact, thereby significantly improving the parking experience comfort for drivers and passengers.

[0031] Energy efficiency optimization and component life extension: By minimizing energy consumption in the objective function, this invention avoids unnecessary excessive torque output, effectively reducing the energy consumption of the parking system. Simultaneously, smooth and precise torque adjustment reduces wear on mechanical components, extending the service life of the parking system.

[0032] Enhanced intelligence and automation: This invention can make optimal decisions autonomously based on complex environmental changes without human intervention, demonstrating a higher level of parking automation and intelligence, and laying a solid foundation for intelligent parking and advanced driver assistance systems for future electric vehicles. Attached Figure Description

[0033] Figure 1 This is a schematic diagram of the overall architecture of the electric vehicle parking motor torque adjustment system of the present invention;

[0034] Figure 2 This is a flowchart of the main process of the electric vehicle parking motor torque adjustment method of the present invention;

[0035] Figure 3 This is a detailed structural diagram of the data acquisition module of the present invention;

[0036] Figure 4 This is a schematic diagram of the vehicle-environment coupled dynamic model construction and parameter update module of the present invention;

[0037] Figure 5 This is a schematic diagram of the adaptive model predictive control strategy execution module of the present invention. Detailed Implementation

[0038] The various exemplary embodiments, features, and aspects of this application will now be described in detail with reference to the accompanying drawings. The same reference numerals in the drawings denote elements that have the same or similar functions. Although various aspects of the embodiments are shown in the drawings, they are not necessarily drawn to scale unless specifically indicated otherwise.

[0039] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments.

[0040] Furthermore, to better illustrate this application, numerous specific details are provided in the following detailed description. Those skilled in the art should understand that this application can be implemented without certain specific details. In some instances, methods, means, components, and circuits well-known to those skilled in the art have not been described in detail in order to highlight the main points of this application.

[0041] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0042] This invention provides a method for adjusting the torque of a parking motor in an electric vehicle, such as... Figures 1 to 5 As shown, this method aims to achieve optimal torque management for electric vehicles in various parking scenarios by constructing an intelligent control framework capable of real-time perception, prediction, and adaptive adjustment. The core idea of ​​this method is to integrate multi-source information, establish and dynamically update a vehicle-environment coupling model, and then utilize an adaptive model predictive control strategy to proactively calculate the optimal torque command for the parking motor, thereby achieving high precision, high safety, high stability, and high energy efficiency in the parking process.

[0043] In one specific implementation, the entire logic flow of electric vehicle parking motor torque regulation begins with the data acquisition phase. This phase forms the perception foundation of the entire control system, continuously acquiring multi-dimensional, real-time data streams from various subsystems and installed sensors within the electric vehicle. For example, vehicles are typically equipped with an inertial measurement unit (IMU), which can output high-frequency triaxial acceleration information (including longitudinal, lateral, and vertical acceleration) and triaxial angular velocity information (including pitch, roll, and yaw rates). This raw data, after appropriate processing and filtering, can be used to accurately estimate the vehicle's real-time attitude, such as pitch and roll angles, which is crucial for determining the road surface slope angle. Simultaneously, each wheel is typically equipped with a wheel speed sensor to monitor the rotational speed of each wheel in real time. By comparing the rotational speeds of different wheels, or combining this with the overall vehicle motion state, it is possible to infer whether the vehicle has a tendency to slip, which plays a key role in estimating the road adhesion coefficient and assessing parking safety. In addition, the pressure sensors inside the braking system provide information on the current hydraulic or air pressure in the brake lines, which is directly related to the actual braking force applied to the brake disc or brake drum, providing an important input for precise control of the parking force.

[0044] In addition to the sensors that directly measure vehicle motion and attitude, the method of this invention also utilizes the vehicle's internal Controller Area Network (CAN) bus to obtain deeper system status information. For example, the Battery Management System (BMS) provides information such as the battery's state of charge (SOC), current, voltage, and temperature via the CAN bus. This data is crucial for assessing the energy consumption of the parking system, optimizing energy efficiency, and ensuring battery safety. Simultaneously, the Drive Motor Controller (DMC) also reports the drive motor's output torque, speed, and operating mode in real time via the CAN bus. This data helps coordinate the interaction between the parking motor and the drive motor when the parking brake is released, ensuring a smooth vehicle start. Regarding environmental perception data, in addition to the initial slope information provided by the IMU, this invention can also utilize data from advanced driver assistance system (ADAS) sensors such as onboard cameras and / or millimeter-wave radar for further analysis. Onboard cameras, through image processing and computer vision algorithms, can identify road texture, lane markings, road signs, and surrounding stationary or moving objects, thereby further refining the perception of road conditions (e.g., wet, dry, gravel roads) and slope. Millimeter-wave radar can provide high-precision distance and relative speed information, helping to detect obstacles in front of or behind the vehicle, and in some scenarios, assisting in slope estimation. Furthermore, combined with high-precision map data, it can provide accurate elevation information of the vehicle's location and preset road slope information, further enhancing the robustness and accuracy of slope angle estimation.

[0045] Following this is the data preprocessing and feature extraction stage. All raw data acquired from different sensors and the CAN bus, due to variations in sampling frequency, data format, units, and potential noise, must undergo rigorous preprocessing before direct use. The first step is noise filtering. For example, high-frequency noise and random jitter in IMU data can be processed using moving average filtering, median filtering, or more complex algorithms such as Kalman filtering and extended Kalman filtering to obtain smooth and accurate vehicle attitude and kinematic data. Time synchronization is another crucial step, as data from different sensors may be acquired at different timestamps. Timestamp alignment algorithms must be used to synchronize all data onto a unified timeline to ensure that all data corresponds to the vehicle's actual state at the same moment in subsequent calculations and model inputs. Unit conversion is also essential, such as converting the raw pulse signals from wheel speed sensors into actual wheel angular or linear velocities, and converting braking pressure into actual force values.

[0046] At this stage, key environmental state estimation and vehicle parameter identification are also performed. For example, slope angle estimation utilizes multi-source fusion, combining pitch angle information from the IMU, altitude change rate data from GPS, and road slope information pre-stored in high-precision maps. Through weighted fusion, Kalman filtering, or other sensor fusion algorithms, the shortcomings of single sensors in complex environments (such as bumpy roads or areas with unstable GPS signals) can be effectively compensated for, improving the accuracy and robustness of slope estimation. Estimating the road surface adhesion coefficient is a challenge in parking force control. This invention can achieve this through the following methods: One method is based on real-time correlation analysis between tire slip ratio and drive / brake torque. When the vehicle applies drive or brake torque, the road surface adhesion coefficient is identified online by monitoring the changing trend of wheel slip ratio and combining it with known tire models. Another method utilizes the vehicle's vibration characteristics and suspension dynamic response, indirectly estimating the road surface adhesion coefficient by analyzing the vibration spectrum or suspension displacement data when the vehicle is traveling on different road surfaces. Furthermore, estimating the total vehicle mass is crucial for parking force calculation, as changes in vehicle load significantly affect the force required for parking. The total mass of a vehicle can be identified online using various methods. For example, during vehicle start-up or acceleration, the output torque of the drive motor and the vehicle's acceleration can be measured, and then inversely calculated using the vehicle's dynamics equations. Alternatively, if the vehicle is equipped with load sensors, the sensor data can be directly used for updates. The real-time and accurate estimation of these parameters provides a solid data foundation for subsequent dynamic model construction and control strategy execution.

[0047] The project proceeds to the stage of constructing and updating the vehicle-environment coupled dynamic model. This invention establishes a multivariable, nonlinear dynamic model designed to comprehensively describe the longitudinal and lateral dynamic behavior of electric vehicles during parking or release, fully considering the transmission characteristics of the parking mechanism and the influence of environmental factors (such as slope and road adhesion coefficient) on vehicle motion. The core of this model lies in the adaptability of its parameters; that is, the key parameters are not fixed but dynamically updated based on real-time estimation results to ensure that the model always closely matches actual vehicle and environmental conditions.

[0048] The model mainly comprises the following sub-modules: a vehicle dynamics model, which describes the vehicle's motion response under various forces (such as parking force, gravity, rolling resistance, and air resistance), including longitudinal acceleration, vehicle speed, and potential lateral slippage. A parking mechanism model, which details how the parking motor's output torque is converted into the actual parking force applied to the wheels through a reduction mechanism, considering nonlinear characteristics such as transmission efficiency, clearance, and friction. An environmental model, which incorporates real-time estimated environmental parameters such as slope angle, road adhesion coefficient, and wind speed into the vehicle dynamics calculations, enabling the model to accurately reflect the vehicle's force under different environmental conditions.

[0049] The most important feature of this model is its online, real-time parameter updates. For example, the total mass of the vehicle... The rolling resistance coefficient varies with the number of passengers and the cargo load; the rolling resistance coefficient is affected by road surface type and temperature; the air resistance coefficient is affected by vehicle speed and air density; and the actual transmission efficiency of the parking mechanism may fluctuate slightly due to wear or temperature changes. This invention utilizes advanced system identification and state estimation algorithms, such as Recursive Least Squares (RLS), Extended Kalman Filter (EKF), or Unscented Kalman Filter (UKF), to iteratively correct the aforementioned key parameters by combining the differences between the actual vehicle motion response (such as vehicle speed and acceleration) and the model's predicted output. In this way, the system always operates on a dynamic model that highly matches the actual vehicle and environment, thereby ensuring the accuracy of predicting the vehicle's future state and providing a reliable foundation for adaptive model predictive control.

[0050] Here, the longitudinal motion equation of the vehicle under parking force is a core component of the vehicle-environment coupled dynamic model of this invention and forms the basis for predicting vehicle dynamic behavior. This equation can be expressed as:

[0051]

[0052] in,

[0053] Indicates the current moment The estimated total mass of the electric vehicle. This is a dynamic parameter that is updated in real time based on changes in the vehicle's load (such as the number of occupants and the weight of cargo) to accurately reflect the vehicle's inertial characteristics.

[0054] Indicates the current moment The vehicle's longitudinal acceleration. Under stable parking conditions, this value approaches zero. During parking or releasing the parking brake, it describes the trend of vehicle speed change.

[0055] Indicates the parking agency at the current moment The equivalent parking force applied to the wheels. This is the actual braking force applied to the wheels after the torque output by the parking motor is converted by the reduction mechanism and the braking system. It is the core variable that this method needs to precisely control.

[0056] This represents the gravitational acceleration constant, typically taking a value of approximately... .

[0057] Indicates the current moment The estimated road slope angle. Positive values ​​are used for uphill sections, and negative values ​​are used for downhill sections. This parameter reflects the component of gravity in the longitudinal direction of the vehicle and has a decisive impact on the parking force requirement.

[0058] Indicates the current moment The rolling resistance experienced by the vehicle due to road surface conditions. This resistance is related to road surface type, tire characteristics, vehicle speed, and vehicle load. This invention dynamically calculates this value based on the real-time estimated road adhesion coefficient and vehicle speed.

[0059] Indicates the current moment Air resistance experienced by a vehicle. This resistance is mainly proportional to the vehicle's frontal area, drag coefficient, and the square of its speed. It is usually small when the vehicle is parked at low speed or stationary, but becomes significant when the vehicle is released from the parking space and begins to move.

[0060] The above equations describe how a vehicle, under parking force, overcomes or is constrained by the gravity component of the slope, road rolling resistance, and air resistance, thus generating longitudinal motion. By updating various parameters online, the model can accurately predict the dynamic response of the vehicle under various complex operating conditions.

[0061] The next stage is determining the target parking force requirement. This stage involves calculating the ideal parking force required to maintain the vehicle's stationary or stable movement based on the driver's intentions and the real-time status of the vehicle and environment. When the driver issues a parking request through various means (such as pressing the parking button, shifting the gear lever into P, or receiving a command from the automatic parking system), the system responds immediately. Upon receiving the parking request, the system will determine the parking force based on the currently estimated slope angle. Based on the road surface adhesion coefficient and a vehicle-environment coupled dynamic model, the minimum parking force required to overcome the current slope's gravity component and road rolling resistance is calculated. For example, in an uphill scenario, the minimum parking force must be sufficient to counteract the vehicle's tendency to slide backward due to gravity; in a downhill scenario, it must resist the vehicle's tendency to slide forward. To maximize parking safety, this invention typically adds a preset safety margin to the calculated minimum parking force. This safety margin is an additional braking force designed to address uncertainties such as model estimation errors, sudden changes in road conditions, or unexpected external impacts, ensuring the vehicle remains firmly parked even in extreme conditions. Ultimately, the minimum parking force plus the safety margin constitutes the target parking force at this point. .

[0062] When the driver requests to release the parking brake, the logic for determining the target parking force is completely different. Instead of immediately reducing the target parking force to zero, it gradually decreases from its current value to zero according to a preset, gradual curve. This progressive reduction strategy aims to ensure a smooth vehicle start, avoiding the jerking or shock that may occur with traditional parking systems during release. The system predicts the vehicle's start-up smoothness at different torque reduction rates based on a vehicle-environment coupled dynamic model and selects the optimal decay curve. For example, when releasing the parking brake downhill, to avoid sudden acceleration due to gravity, the system may maintain a weak auxiliary braking force until the driver applies sufficient driving torque to overcome the slope resistance and the vehicle begins to move steadily forward before fully releasing the parking force. This refined target parking force determination mechanism is key to achieving the high smoothness of parking and release in this invention.

[0063] In the core adaptive model predictive control (AMPC) strategy execution phase, this invention utilizes the vehicle-environment coupled dynamic model updated in the previous phase and the determined target parking force requirement to solve an optimization problem online, in order to determine the future time period (i.e., the prediction time domain). The optimal torque command sequence for the parking motor. This optimization process is performed in each control cycle. The process will be repeated, demonstrating the "adaptive" nature of the controller.

[0064] The basic idea of ​​AMPC is that, within each control cycle, the controller uses the most accurate vehicle-environment coupled dynamic model to predict the possible motion states of the vehicle at multiple future time steps. Based on this, an optimization algorithm calculates the possible motion states of the vehicle in the finite control time domain in the future. A series of parking motor torque commands are generated to make the predicted state of the vehicle as close as possible to the target parking state (e.g., zero longitudinal speed and no slippage), while also meeting a series of performance indicators and safety constraints. These performance indicators typically include: minimizing the deviation between the parking state and the target state (i.e., achieving high-precision parking), minimizing the rate of change of control variables (i.e., ensuring smooth torque output and reducing shocks), and minimizing energy consumption.

[0065] Specifically, the objective function used for online optimization in this invention can be written as:

[0066]

[0067] in,

[0068] At the current moment The cost function to be minimized. The smaller its value, the closer the predicted parking performance is to the ideal state.

[0069] Indicates the current moment Future moments predicted based on the updated vehicle-environment coupled dynamic model The vehicle state vector. This vector typically contains several key kinematic and dynamic states of the vehicle, such as the vehicle's longitudinal velocity, longitudinal acceleration, the state of the parking mechanism (e.g., the relative position or pressure of the brake disc and brake pads), and other internal state variables that may affect the parking force.

[0070] It is the corresponding reference target state vector, which defines the vehicle at future moments. The desired ideal parking state. In a fully parked scenario, this typically means that the vehicle's longitudinal velocity and acceleration are both zero, and the parking mechanism is in a fully loaded state to counteract external disturbances.

[0071] It is the weighted matrix of state errors, which is a positive definite or positive semi-definite matrix used to measure the predicted state vector. With reference target state vector Penalty for deviations between them. By adjusting The values ​​of different diagonal elements in the matrix can balance the tracking accuracy of different state variables (such as velocity error and acceleration error), giving higher weight to more important states.

[0072] Indicates future time The change in the parking motor torque command, i.e. This feature aims to penalize drastic changes in parking motor torque commands to ensure smooth control during parking, reducing mechanical shock and user discomfort.

[0073] It is a weighted matrix of the rate of change of the control quantity; it is also a positive definite or semi-positive definite matrix, used to penalize torque commands. Rapid changes. By adjusting The elements of the matrix can be used to strike a trade-off between control response speed and output stability. Larger... The value will prompt the controller to generate a smoother sequence of torque commands.

[0074] It is the length of the prediction time domain, representing the number of time steps the controller takes to predict the future vehicle state.

[0075] This refers to the length of the control time domain, representing the number of time steps the controller takes to calculate the optimal torque command for the future parking motor. Typically, the control time domain... Less than or equal to the prediction time domain .

[0076] This optimization process is typically achieved by solving a quadratic programming (QP) problem. This problem not only includes the objective function described above but is also subject to a series of stringent constraints. These constraints include the maximum / minimum torque output limits of the parking motor itself, physical limitations on the rate of torque change (to prevent motor overload or excessively rapid response causing shock), and safety constraints such as the maximum permissible slip distance of the vehicle during parking. For example, when parking on a slope, the system constrains the parking force to be sufficient to overcome the gravitational component and rolling resistance, and the vehicle's slip distance to be kept within a very small safety range. AMPC's "adaptive" characteristic is reflected in the fact that before each optimization, the parameters of the vehicle-environment coupled dynamic model are updated online based on the latest observation data, ensuring that the controller always makes predictions and decisions based on the most accurate vehicle dynamic characteristics. This dynamic adjustment ensures that the controller can provide optimal torque commands under different operating conditions.

[0077] In practical applications, a crucial transmission ratio exists between the torque output by the parking motor and the parking force of the vehicle. Although the optimization objective of AMPC is usually to directly generate the optimal parking motor torque command, understanding and accurately grasping the conversion relationship between parking force and motor torque is essential when designing control strategies. This relationship can be approximated as:

[0078]

[0079] in, This represents the torque that the parking motor needs to output during the current control cycle. This is the direct output of the AMPC strategy, used to drive the parking motor. This represents the expected equivalent parking force to be applied to the wheels. This force originates from the target parking force requirement and represents the actual force needed to stop the vehicle from moving. Within the AMPC, this expected force is used to guide the optimization of motor torque through reverse calculation or as part of constraints. This is the effective radius of the wheel. It converts the force on the wheel into torque, or the torque of the wheel into force, and is an important geometric parameter in mechanical conversion. This refers to the reduction ratio of the parking mechanism. The parking motor typically converts its high-speed, low-torque motor output into low-speed, high-torque braking force through a multi-stage gear or worm gear reduction mechanism. This reduction ratio determines the torque amplification factor. This refers to the transmission efficiency of the parking mechanism. In actual physical transmission processes, due to factors such as friction and energy loss, the input torque is not completely converted into the output torque. The transmission efficiency is a coefficient less than 1, reflecting the effectiveness of energy conversion.

[0080] In actual operation, the reduction ratio of the parking mechanism It is usually a fixed value, but its transmission efficiency The parameters may fluctuate slightly due to mechanical wear, temperature changes, lubrication conditions, and load magnitude. This invention ensures the accuracy of the conversion from parking force to motor torque at all times by including real-time estimation and updating of these parameters in the vehicle-environment coupled dynamic model. This guarantees that the optimal torque command calculated by the AMPC can accurately achieve the required parking force. Although this formula is not a direct calculation step when the AMPC outputs motor torque, it describes the physical characteristics of the parking mechanism and serves as the fundamental physical constraint and knowledge for prediction and optimization within the AMPC internal model.

[0081] Once the optimal torque command sequence is available Once calculated, the optimal torque command for the current control cycle will be sent during the parking motor drive phase. The signal is sent to the parking motor controller. The parking motor controller is the core of the parking system, responsible for translating the abstract torque commands from the higher-level control unit (AMPC) into actual motor drive signals. Specifically, the controller employs appropriate advanced control algorithms, such as current closed-loop control, field-oriented control (FOC), or direct torque control (DTC), depending on the motor type (e.g., stepper motor, brushless DC motor, or permanent magnet synchronous motor). These algorithms precisely regulate the current and voltage flowing through the motor windings, thereby driving the parking motor to output the required torque precisely.

[0082] The parking motor is typically a small, high-power-density motor whose output shaft is connected to the vehicle's braking system via a high-precision reduction gear. This reduction gear converts the motor's high-speed, low-torque rotational motion into low-speed, high-torque braking action to meet parking force requirements. The reduction gear may employ various forms such as planetary gears, worm gears, or lead screws, designed to provide sufficient reduction ratio and transmission efficiency while ensuring a compact structure and good durability. The reduced torque is ultimately transmitted to the braking system, acting on brake discs or drums, clamping the friction pads or pushing the brake shoes to apply sufficient braking force to the wheels, thus parking the vehicle. The response speed and precision of the entire drive chain directly affect the effectiveness of the AMPC strategy and the smoothness of the final parking experience.

[0083] Throughout the parking and release process, performance evaluation and model updates are ongoing. This is a closed-loop feedback mechanism designed to continuously monitor the parking system's performance and adjust and optimize model parameters as needed to address changes in actual operating conditions. The system monitors key parking states in real time. For example, high-precision wheel speed sensors continuously monitor for minor slippage on each wheel (even in the parking state, minor slippage may occur due to insufficient road adhesion coefficient or errors in parking force calculation). Simultaneously, acceleration sensors on the vehicle chassis or brake calipers monitor for any impacts or unnecessary vibrations during parking or release.

[0084] If there is a significant deviation between the actual parking effect (such as vehicle slippage and starting impact) and the target state predicted by AMPC, it indicates that some parameters of the vehicle-environment coupled dynamic model may have large errors, or that the external environment (such as the road surface adhesion coefficient) has changed drastically without being accurately perceived. In this case, the system will trigger further online re-identification and updating of model parameters. For example, if slight slippage is detected when the vehicle is parked on a slope, the system will infer that the current estimate of the road surface adhesion coefficient may be too high, and then adjust the estimation algorithm or its output value. Similarly, if there is a noticeable jerk when the vehicle starts, the system may fine-tune the estimate of the parking mechanism's transmission efficiency or correct the estimation algorithm for the vehicle's total mass to improve the model's prediction accuracy in the next control cycle. This continuous feedback, error detection, and adaptive adjustment mechanism ensures that the method of this invention maintains excellent performance under various dynamic conditions, effectively improving the system's robustness and long-term adaptability.

[0085] Furthermore, this invention fully demonstrates its advantages in fine-tuning during the parking release process. Traditional parking systems often release braking force abruptly upon release, causing a noticeable jerk in the vehicle, especially on inclines. The AMPC system of this invention, however, employs a more intelligent and smooth release strategy. When the driver requests to release the parking brake, the AMPC system does not immediately reduce the parking torque to zero. Instead, it uses the latest vehicle-environment coupled dynamic model to proactively predict the vehicle's starting smoothness at different torque reduction rates. Based on these predictions, the system calculates an optimal torque decay curve. This curve ensures that the parking force gradually decreases smoothly and without impact from its current value until it is completely released within a set time, thus guaranteeing a highly smooth start when the driver depresses the accelerator pedal. For example, in the special condition of starting on a downhill slope, to prevent the vehicle from suddenly accelerating due to gravity after the parking force is fully released, the system may maintain a weak braking torque during the parking release process, allowing the vehicle to move only in a controlled, slow coasting state, until the driver applies sufficient driving torque to overcome the slope resistance and the vehicle begins to accelerate steadily before finally releasing the parking force completely. This "hill start" or gradual release strategy significantly improves the comfort of the driver and passengers during the parking release process, avoiding the discomfort caused by traditional parking systems.

[0086] The method of this invention demonstrates strong versatility and excellent performance in multiple application scenarios:

[0087] Scenario 1: Parking on a flat road. Under this ideal condition, the road slope angle is... With the torque close to zero, the gravitational component affecting the vehicle is minimal. At this point, the vehicle-environment coupled dynamic model identifies the primary force to overcome as road rolling resistance. The AMPC strategy calculates a small, optimal parking motor torque command based on the vehicle's current state and a preset safety margin. This torque effectively ensures stable parking while minimizing energy consumption in the parking system and avoiding unnecessary overload. Upon releasing the parking brake, AMPC smoothly releases the parking torque to zero according to a preset, gradual torque decay curve, ensuring a smooth and seamless start-up process on a flat road surface.

[0088] Scenario 2: Parking on an uphill slope. When the vehicle is on an uphill road, the slope angle... When the value is positive, the vehicle will experience a rearward gravitational component, causing it to tend to slide backward. In the method of this invention, the data acquisition module and data preprocessing module accurately estimate the current uphill slope angle. Based on this precise slope angle, the vehicle-environment coupled dynamic model calculates the minimum parking force sufficient to overcome the rearward gravitational component and rolling resistance. The AMPC strategy adds a safety margin to this minimum parking force and outputs a larger parking motor torque command. Throughout the uphill parking process, the performance evaluation and model update module continuously monitors the wheel speed sensor data. Once any slight backward slippage trend is detected, AMPC responds immediately, fine-tuning the parking motor torque to ensure the vehicle remains firmly stationary on the uphill road with a millisecond-level response speed, preventing accidental slippage.

[0089] Scenario 3: Parking on a downhill slope. When the vehicle is on a downhill road, the slope angle... A negative value indicates that the vehicle will experience a forward-rolling component of gravity, causing it to tend to slide forward. Similar to uphill parking, the system accurately estimates the downhill slope angle. The vehicle-environment coupled dynamic model calculates a parking force sufficient to counteract the forward gravity component and rolling resistance. The AMPC strategy outputs the corresponding parking motor torque based on this requirement. Downhill parking release is a highlight of this invention. To avoid sudden acceleration due to gravity after the parking force is released, the AMPC system pays special attention to the smooth release of torque. It may employ a "hill start" strategy, where the parking force is not immediately and completely released upon release. Instead, as the vehicle begins to glide forward slowly but remains under control, the parking force is gradually reduced based on the driver's driving needs and the slope, until the driver's applied driving torque is sufficient to stably control the vehicle's forward acceleration. This ensures the safety and smoothness of the vehicle during downhill starts, effectively avoiding the dangers of sudden acceleration.

[0090] Scenario 4: Parking under different loads. When the load on an electric vehicle changes, such as adding passengers or loading cargo, the total mass of the vehicle... This will change accordingly. The total vehicle mass is one of the key parameters affecting parking force calculation, and traditional parking systems often cannot dynamically adapt to such changes. This invention, through its online parameter identification mechanism in its vehicle-environment coupled dynamic model, can update the total vehicle mass in real time. The estimated value is obtained by using a reverse dynamics model to accurately estimate the total mass of the vehicle during acceleration from a standstill. For example, during real-time measurement of the drive motor torque and vehicle acceleration, the current total mass is accurately estimated. Furthermore, during AMPC optimization, the controller always uses the most accurate vehicle total mass model for prediction and decision-making, thereby avoiding problems such as insufficient parking force (vehicle slippage) or overloaded parking force (energy waste, component wear) caused by inaccurate mass estimation, ensuring accurate and safe parking under any load conditions.

[0091] This invention also considers a system failure or anomaly handling mechanism, further enhancing the system's robustness and safety. For example, if a critical sensor (such as an IMU or wheel speed sensor) malfunctions, resulting in data loss or anomalies, the system will immediately activate a backup data source (if available) or use a model prediction method to estimate the missing data to maintain basic parking functionality. Simultaneously, the system will issue a clear warning to the driver, indicating the sensor failure and suggesting appropriate action. More importantly, if the AMPC controller, while executing the optimization algorithm, detects that it cannot find a parking motor torque command sequence that fully meets all safety requirements under the currently set constraints (e.g., maximum parking force, minimum release force, maximum permissible slip distance), the system will prioritize parking safety. In this case, the controller will abandon the pursuit of optimal performance and instead immediately apply a preset, rigorously verified maximum safe parking force to prevent accidental vehicle slippage or entry into an unsafe state, and will issue another warning to the driver. This fail-safe mechanism ensures that the vehicle maintains the highest level of parking safety even in the event of system anomalies.

[0092] In summary, the present invention provides a method for regulating the torque of an electric vehicle parking motor. By introducing multi-source information fusion technology to perceive the vehicle and environmental status in real time, a high-precision vehicle-environment coupled dynamic model is constructed and dynamically updated. An adaptive model predictive control strategy is then used for forward-looking torque management, achieving high-precision, highly adaptable, highly safe, highly stable, and highly energy-efficient regulation of the parking motor torque. This method significantly improves the parking performance and user experience of electric vehicles, effectively solving the problems of insufficient parking force control accuracy, lag response, and low energy consumption efficiency in existing technologies under complex operating conditions.

Claims

1. A method for adjusting the torque of a parking motor in an electric vehicle, characterized in that, include: Real-time collection of multi-dimensional vehicle operation status data, driver operation commands, and environmental perception data; Based on real-time collected data, a coupled dynamic model of the current vehicle and environment is established, and the parameters of the coupled dynamic model are updated in real time. The target parking force requirement is determined based on the driver's operating instructions and the coupled dynamic model of the vehicle and the environment. Using an adaptive model predictive control strategy, based on the coupled dynamic model of the vehicle and the environment and the target parking force requirement, the future motion state of the vehicle is predicted within a preset prediction time domain, and the optimal torque command of the parking motor in the current control cycle is optimized and calculated. as well as The optimal torque command is sent to the parking motor controller to drive the parking motor to generate the corresponding torque to enable the vehicle to be parked or released from parking. The process of establishing a coupled dynamic model of the current vehicle and its environment, and updating the parameters of the coupled dynamic model in real time, includes: Establish a coupled dynamic model that includes a vehicle dynamics model, a parking mechanism model, and an environmental model; The vehicle dynamics model describes the longitudinal motion response of the vehicle under the influence of parking force, gravity, rolling resistance, and air resistance. The parking mechanism model describes the transmission characteristics of converting the output torque of the parking motor into a parking force applied to the wheels; and The environmental model incorporates real-time estimated slope angles and road surface adhesion coefficients into vehicle dynamics calculations. The real-time updating of the parameters of the coupled dynamic model includes: Using recursive least squares, extended Kalman filtering, or unscented Kalman filtering system identification and state estimation algorithms, and combining the difference between the actual vehicle motion response and the model prediction output, the total vehicle mass, rolling resistance coefficient, air resistance coefficient, and transmission efficiency of the parking mechanism in the coupled dynamic model are iteratively corrected. The determination of the target parking force requirement includes: Upon receiving a parking request, the system calculates the minimum parking force required to keep the vehicle stationary based on the currently estimated road slope angle and road adhesion coefficient, and adds a preset safety margin to this minimum parking force to determine the target parking force; and When a parking release request is received, the vehicle starts smoothly based on the vehicle-environment coupled dynamic model, and an optimal torque decay curve is calculated so that the target parking force decreases smoothly from the current value to zero according to the optimal torque decay curve.

2. The method for adjusting the torque of the parking motor of an electric vehicle according to claim 1, characterized in that, The real-time acquisition of multi-dimensional vehicle operating status data, driver operation commands, and environmental perception data includes: The vehicle's real-time attitude, kinematic information, and braking pressure information are obtained through sensors inside the vehicle. The system obtains battery state-of-charge data from the battery management system and drive motor torque and speed information from the drive motor controller via the vehicle network; and The road surface slope angle and road surface adhesion coefficient are estimated by using onboard environmental perception sensors and / or high-precision map information.

3. The method for adjusting the torque of the parking motor of an electric vehicle according to claim 2, characterized in that, The acquisition of real-time vehicle attitude, kinematic information, and braking pressure information through in-vehicle sensors includes: The vehicle's three-axis acceleration and three-axis angular velocity information are obtained through the inertial measurement unit, and the vehicle's pitch and roll angles are calculated. The real-time rotational speed of each wheel is obtained through wheel speed sensors; and Hydraulic or pneumatic pressure information in the brake lines is obtained through a pressure sensor in the braking system.

4. The method for adjusting the torque of the parking motor of an electric vehicle according to claim 2, characterized in that, The process of obtaining road surface slope angle estimates and road surface adhesion coefficient estimates through vehicle-mounted environmental perception sensors and / or high-precision map information includes: By comprehensively utilizing pitch angle information from the inertial measurement unit, altitude change rate data from the Global Positioning System, and road slope information pre-stored in high-precision maps, the current road surface slope angle is estimated; and Based on real-time correlation analysis between tire slip ratio and driving / braking torque, or vehicle vibration characteristics and suspension dynamic response, the road adhesion coefficient is estimated online.

5. The method for adjusting the torque of an electric vehicle parking motor according to claim 1, characterized in that, The adaptive model predictive control strategy, which predicts the future motion state of the vehicle and optimizes the calculation of the optimal torque command for the parking motor in the current control cycle, includes: Within each control cycle, the future motion state of the vehicle is predicted in the prediction time domain using the updated coupled dynamic model of the vehicle and the environment. By solving a quadratic programming problem, the optimal torque command sequence of the parking motor in the future control time domain is obtained. The objective function of the quadratic programming problem includes a state tracking error term and a control variable change rate term. The state tracking error term penalizes the deviation between the predicted vehicle state and the target parking state; and The control variable change rate term penalizes drastic changes in the parking motor torque command.

6. The method for adjusting the torque of an electric vehicle parking motor according to claim 5, characterized in that, Solving the quadratic programming problem is also subject to a series of constraints, including: Maximum and minimum torque output limits for the parking motor; The physical limitations of the parking motor torque variation rate; and Safety constraints on the maximum permissible slip distance of a vehicle during parking.

7. The method for adjusting the torque of an electric vehicle parking motor according to claim 1, characterized in that, Sending the optimal torque command to the parking motor controller to drive the parking motor to generate corresponding torque to achieve vehicle parking or parking release, also includes: The parking motor controller drives the parking motor using a current closed-loop control or field-oriented control algorithm based on the optimal torque command; and The system monitors the parking status in real time, including the vehicle's slippage and impact conditions. If there is a significant deviation between the actual parking effect and the predicted status, it triggers online identification and updating of the parameters of the coupled dynamic model of the vehicle and the environment to improve the model's prediction accuracy.

Images