An adaptive optimization new energy vehicle wheel hub motor efficiency control method

Abstract

The present application belongs to the field of new energy vehicle power transmission and drive control technology, and particularly relates to a self-adaptive and optimized new energy vehicle wheel hub motor efficiency control method, which comprises the following steps: obtaining a state signal of a new energy vehicle wheel hub motor; evaluating a real-time matching relationship between motor input power and vehicle mechanical kinetic energy change rate based on the state signal to obtain an electromechanical energy decoupling index; calculating a road mutation confidence index based on the time domain evolution characteristics of the electromechanical energy decoupling index and in combination with the frequency domain response of wheel kinematics; obtaining a transient correction boundary in combination with real-time slip energy lost in driving wheel idling at the current moment; substituting the transient correction boundary into a cost function of an adaptive model as a constraint condition to solve an optimal electromagnetic torque sequence, with the first term serving as an instruction to be issued to a driver to control the motor to output the torque, thereby realizing self-adaptive efficiency control. The present application enhances real-time road mutation sensing capability and prolongs vehicle range.

Description

Technical Field

[0001] This invention relates to the field of powertrain and drive control technology for new energy vehicles. More specifically, this invention relates to an adaptive optimization method for energy efficiency control of hub motors in new energy vehicles. Background Technology

[0002] With the advancement of green transformation and intelligent upgrading of global industrial and mining enterprises, hub motor independent drive technology has become the core technology path for the electrification development of heavy-duty construction machinery due to its significant advantages such as short transmission chain, high space utilization, and independent torque control. In scenarios such as mining, port operations, and high-intensity infrastructure construction, vehicles often face extremely harsh and variable road conditions. Due to the huge load, high center of gravity, and strong transient torque burst force at the wheel end of heavy-duty vehicles, maintaining efficient power coupling between the wheels and the ground is not only directly related to work efficiency, but also involves the energy consumption control of the whole vehicle and the service life of the drive system.

[0003] Currently, to achieve optimal torque distribution and anti-skid control on complex and rugged road surfaces, the industry typically employs adaptive model predictive control (AMDC) technology. This technology identifies parameters such as the road adhesion coefficient online, updates the vehicle dynamics model, and predicts future states based on the vehicle dynamics model to optimize control commands. However, existing adaptive parameter identification algorithms often rely on the statistical characteristics of historical data, resulting in inherent convergence lag. When a construction vehicle instantly moves from a hard rock surface into a soft, muddy area, the parameter identifier cannot lock onto a new low adhesion coefficient within milliseconds. Within this identification blind zone, the MDC still calculates torque based on the old high adhesion coefficient model, leading to excessive motor output, causing severe wheel slippage and spinning, resulting in significant energy waste, increased tire wear, and deterioration of vehicle stability. Sudden longitudinal slippage may induce yaw instability or loss of steering control, threatening operational safety. Furthermore, energy efficiency drops sharply, with a large amount of electrical energy being converted into useless heat and kinetic energy, significantly shortening the driving range. Summary of the Invention

[0004] To address the aforementioned technical problems of transient control blind spots, dynamic coupling imbalance, and overall energy efficiency failure caused by parameter identification convergence lag, this invention provides an adaptive optimization method for energy efficiency control of in-wheel motors in new energy vehicles. The method includes: acquiring operating status data of the in-wheel motor at current and historical time points, and performing noise reduction processing on the operating status data to obtain a high-confidence status signal; evaluating the real-time matching relationship between the motor input power and the rate of change of vehicle mechanical kinetic energy based on the status signal to obtain an electromechanical energy decoupling index; and based on the time-domain evolution characteristics of the electromechanical energy decoupling index, combining the frequency domain of wheel kinematics... In response, a pavement abrupt change confidence index is calculated to assess the confidence level that the dynamic anomaly is caused by a sudden change in the actual pavement adhesion performance. Based on the pavement abrupt change confidence index, combined with the real-time slip energy lost in the drive wheel idling at the current moment, a transient correction boundary is obtained to limit the upper limit of the model predictive control solution. The transient correction boundary is substituted as a time-varying hard constraint into the cost function of the adaptive model predictive control to solve for the optimal electromagnetic torque sequence that satisfies the ground adhesion limit. The first term of the optimal electromagnetic torque sequence is sent as a command to the hub motor driver, and the driver uses closed-loop control to output the torque of the motor, thereby realizing adaptive energy efficiency control.

[0005] This invention constructs an electromechanical energy decoupling index by calculating the matching relationship between the motor input power and the rate of change of vehicle mechanical kinetic energy in real time, accurately identifying road surface abrupt changes. Furthermore, it integrates the road surface abrupt change confidence index with real-time slip energy to dynamically generate a transient correction boundary, effectively distinguishing between sudden changes in actual road surface adhesion and random mechanical impacts, avoiding control erroneous triggering. This boundary is used as a hard constraint for model predictive control, enabling the system to immediately tighten the torque upper limit at the moment of a road surface step change, eliminating the identification blind spot caused by parameter identification lag, significantly reducing the ineffective energy consumption caused by excessive torque output from the motor on low-adhesion roads, reducing abnormal tire wear, improving the vehicle's ride smoothness and stability under complex and rugged road conditions, and extending the driving range on a single charge. This provides a safe and reliable drive control solution for new energy heavy-duty vehicles in high-risk working conditions such as mining and port transportation.

[0006] Preferably, the operating status data includes motor electromagnetic torque, real-time wheel angular velocity, vehicle longitudinal acceleration, wheel end vertical load, and vehicle speed sensing signal; real-time wheel angular acceleration and vehicle longitudinal centerline velocity; and the equivalent translational mass of a single wheel, the moment of inertia of the wheel and its transmission components, the wheel rolling radius, and the equivalent internal resistance coefficient of the motor, all pre-stored in the control unit.

[0007] Preferably, the electromechanical energy decoupling index is calculated as follows: In the formula, For electromechanical energy decoupling indicators; This represents the current real-time electromagnetic input power of the motor. The equivalent translational mass of a single wheel; The speed along the longitudinal centerline of the vehicle; This refers to the vehicle's actual longitudinal acceleration. The moment of inertia of the wheel and its transmission components; This refers to the real-time angular velocity of the wheel. This refers to the real-time angular acceleration of the wheel; This represents the average input power of the motor at the current moment. It is a very small constant.

[0008] This invention obtains electromechanical energy decoupling indicators by calculating the deviation between the current real-time electromagnetic input power of the motor and the rate of change of the vehicle's longitudinal and rotational kinetic energy. It clearly presents the loss of driving energy during the electromechanical conversion process, providing data support for the accurate identification of subsequent road conditions and reducing energy efficiency assessment errors caused by sensor noise interference.

[0009] Preferably, the formula for calculating the road surface abrupt change confidence index is: In the formula, The confidence index for road surface abrupt changes; This is a road surface feature normalization factor. The length of the sliding time window; The rate of change of the electromechanical energy decoupling index at the i-th sampling point in the sliding time window; This is the transient impact characteristic of the electromechanical energy decoupling index at the i-th sampling point in the sliding window.

[0010] This invention calculates the road surface change confidence index based on the rate of change of the electromechanical energy decoupling index within the sliding time window and its transient impact characteristics. By capturing the high-frequency jumps of the index in the time domain, the system can quickly respond to the instantaneous decrease in road surface adhesion strength, reducing the hysteresis phenomenon of conventional control algorithms when facing road surface property changes and ensuring the timeliness of hub motor torque adjustment.

[0011] Preferably, the formula for calculating the transient correction boundary is: In the formula, For transient boundary correction; C represents the conventional identification value estimated by the recursive least squares method; C is the confidence index for road surface abrupt change. The real-time slip energy at the current moment; The maximum allowable slip energy threshold; This is the attenuation gain coefficient. This is a factor contributing to the persistence of road surface anomalies.

[0012] This invention utilizes the road surface change confidence index to dynamically correct the conventional identification value estimated by the recursive least squares method, and introduces the real-time slip energy lost in the drive wheel idling to obtain the transient correction boundary. When the slip energy is detected to exceed the maximum allowable slip energy threshold, the torque upper limit is reduced by attenuating the gain coefficient, thereby reducing the risk of large-scale idling of the drive wheel on wet and slippery road surfaces.

[0013] Preferably, the real-time slip energy at the current moment is equal to the product of the motor electromagnetic torque and the wheel angular velocity minus the product of the ground adhesion force and the vehicle speed.

[0014] Preferably, the cost function is calculated as follows: In the formula, Let be the cost function to be minimized; N is the prediction time domain length in model predictive control. To predict the first in the time domain Real-time slip rate of the step; The target slip ratio; Let K be the electromagnetic torque sequence of the k-th motor to be solved. The equivalent internal resistance coefficient of the motor; and Preset weighting coefficients; The preset reference torque; This is the preset reference internal resistance coefficient.

[0015] This invention comprehensively considers the real-time slip ratio deviation in the prediction time domain and the internal resistance loss generated by the electromagnetic torque of the motor in the cost function. By balancing driving stability and energy consumption control through preset weight coefficients, it minimizes motor winding losses while ensuring that the vehicle tracks the target slip ratio, thereby reducing heat accumulation in the hub motor during long-term operation.

[0016] Preferably, the constraint condition is: In the formula, To predict the optimal electromagnetic torque sequence of the motor to be solved in the time domain; Correct the boundary for transient attachment; The real-time vertical load at the wheel end is measured by the sensor; This is the wheel's rolling radius.

[0017] Preferably, the preset weighting coefficient satisfies And require .

[0018] Preferably, the noise reduction method refers to Kalman filtering.

[0019] The beneficial effects of this invention are as follows: This invention constructs an electromechanical energy decoupling index by calculating the matching relationship between the motor input power and the rate of change of vehicle mechanical kinetic energy in real time, accurately identifying road surface abrupt changes. Furthermore, it integrates the road surface abrupt change confidence index with real-time slip energy to dynamically generate a transient correction boundary, effectively distinguishing between sudden changes in actual road surface adhesion and random mechanical impacts, avoiding control erroneous triggering. This boundary is used as a hard constraint for model predictive control, enabling the system to immediately tighten the torque upper limit at the moment of a road surface step change, eliminating the identification blind spot caused by parameter identification lag, significantly reducing the ineffective energy consumption caused by excessive torque output from the motor on low-adhesion roads, reducing abnormal tire wear, improving the vehicle's ride smoothness and stability under complex and rugged road conditions, and extending the driving range on a single charge. This provides a safe and reliable drive control solution for new energy heavy-duty vehicles in high-risk working conditions such as mining and port transportation. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating an adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to the present invention. Figure 2 This is a schematic diagram showing the comparison of wheel end linear velocity response; Figure 3 This is a schematic diagram showing the comparison of motor output torque commands; Figure 4 This is a schematic diagram illustrating the cumulative comparison of ineffective energy loss. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0023] This invention discloses an adaptive optimization method for energy efficiency control of hub motors in new energy vehicles, referring to... Figure 1 This includes steps S1 to S5: S1: Real-time acquisition of multi-source data, noise reduction preprocessing, and output of high-confidence status signals.

[0024] It should be noted that during the operation of the hub motor, sensor data is often mixed with electromagnetic interference and mechanical vibration noise, which directly affects the reliability of subsequent analysis. If the raw data is used directly, the system will have difficulty accurately identifying changes in road surface condition, leading to an increased risk of misjudgment. This step uses data denoising to filter out random noise interference and extract high-confidence state signals, providing a stable and reliable data input basis for subsequent road surface condition assessment, and ensuring that the system's perception of road surface changes is not interfered with by noise.

[0025] This step utilizes the high-precision sensors built into the wheel hub motor of the new energy vehicle and the inertial measurement unit on the vehicle to acquire the operating status data of the wheel hub motor at each sampling moment, both currently and at multiple historical moments. Specifically, the system acquires and processes the following physical quantities according to their attributes: (1) Real-time sensor data acquisition: motor electromagnetic torque, wheel real-time angular velocity, vehicle longitudinal acceleration, wheel end vertical load, and vehicle speed sensing signal used for fusion calculation.

[0026] (2) Kinematics-derived quantities: Based on high-frequency sampling data, the real-time angular acceleration of the wheel is obtained through differential operation, and the longitudinal centerline velocity of the vehicle is calculated by combining the multi-wheel speed fusion algorithm.

[0027] (3) System inherent parameters: call the equivalent translational mass of a single wheel, the moment of inertia of the wheel and its transmission components, the rolling radius of the wheel and the equivalent internal resistance coefficient of the motor, which are pre-stored in the control unit.

[0028] In addition, considering the complex electromagnetic environment at the construction site, the collected operating status data is often accompanied by high-frequency noise. In this embodiment, Kalman filtering is used to preprocess the above operating status data and output a high-confidence status signal, so that the system can dynamically adapt to the sensor noise fluctuations caused by different road bump intensities.

[0029] S2: Obtain electromechanical energy decoupling index based on high-confidence state signals.

[0030] It should be noted that in the actual electromechanical conversion process, there is still a fuzzy state of pseudo-balance in the original energy flow data. That is, the input electrical energy at the motor end and the mechanical response at the wheel end are not directly equivalent in terms of dimensions. This misalignment of dimensions masks the true physical coupling characteristics, making it difficult for the system to directly determine whether the current energy loss is caused by normal mechanical friction or by power leakage due to instability of the drive wheel. In order to distinguish this coupling quality at the energy level, this step further calculates the real-time matching relationship between the motor input power and the rate of change of vehicle mechanical kinetic energy based on high-confidence state signals, and obtains the electromechanical energy decoupling index. The aim is to obtain the degree of nonlinear deviation between the motor input energy and the actual mechanical motion response of the vehicle, as a physical basis for identifying sudden changes in road surface adhesion performance.

[0031] In a stable operating condition, when the wheel end adhesion is sufficient, most of the electromagnetic power output by the motor is converted into the rate of change of translational kinetic energy after the vehicle overcomes resistance and the rate of change of rotational kinetic energy of the wheel itself. In a slipping operating condition, once the road surface adhesion coefficient drops instantly, the effective coupling between the wheel and the ground is lost, and the motor output power cannot be effectively transmitted to the ground to be converted into vehicle acceleration, instead forcing the wheel angular velocity to increase sharply.

[0032] At this point, a significant energy gap will appear between the input power and the effective traction power. Based on the above physical logic, the formula for calculating the electromechanical energy decoupling index is as follows: ; In the formula, It is an electromechanical energy decoupling index used to characterize the degree of nonlinear instability in the energy output of an electromechanical system. The larger the value, the more decoupled the motor energy from the vehicle's motion, resulting in more severe slippage and requiring forced torque limiting. The smaller the value, the tighter the electromechanical coupling, the more stable the driving, and the better the power output can be maintained; This represents the current real-time electromagnetic input power of the motor. The equivalent translational mass of a single wheel is equal to the sum of the translational inertial component of the whole vehicle acting on the drive wheel and the inertial equivalent component of the rotating parts. It is used to relate the wheel-end driving force to the whole vehicle acceleration in the longitudinal dynamic equation. The longitudinal centerline velocity of the vehicle is obtained by fusing multiple wheel speeds. The longitudinal acceleration of the vehicle; The moment of inertia of the wheel and its transmission components; This refers to the real-time angular velocity of the wheel. This refers to the real-time angular acceleration of the wheel; This represents the average input power of the motor at the current moment. It is a very small constant used to avoid numerical singularities caused by the denominator being zero during initial system startup or low speed.

[0033] This electromechanical energy decoupling index, through real-time monitoring of energy flow, can sensitively reflect abrupt changes in road surface characteristics: (1) When the vehicle is in a low stable state, the measured power of the molecular part is highly matched with the mechanical conversion power when the vehicle is accelerating normally or driving at a constant speed. The difference is only a small internal loss, and the electromechanical energy decoupling index is maintained at a stable low level close to 0.

[0034] (2) When the vehicle is in the step-triggered state, when the vehicle moves from the high-adhesion road surface to the low-adhesion area instantly, the maximum friction torque that the ground can provide decreases, which causes the actual longitudinal acceleration of the vehicle to be unable to be maintained with the current real-time electromagnetic input power of the motor. At the same time, the wheel angular acceleration is driven by the residual torque and surges. At this time, the numerator value of the calculation formula will jump exponentially, which will cause the electromechanical energy decoupling index to produce a significant step signal.

[0035] By calculating this electromechanical energy decoupling index, the system can detect the instability trend of the drive wheel in advance through the energy conservation perspective of the imbalance between income and expenditure, without relying on complex road surface estimation models.

[0036] S3: Obtain the confidence index of road surface abrupt change based on electromechanical energy decoupling index.

[0037] It should be noted that in complex heavy-duty truck operation scenarios, there exists a state of physical ambiguity. When a vehicle runs over gravel or travels over bumpy roads and experiences random mechanical impacts, the electromechanical energy flow will also generate short-term pulse fluctuations due to the instantaneous and drastic changes in the vertical load at the wheel ends. This nonlinear fluctuation caused by external vibration is very similar in characteristic shape to the actual road surface adhesion drop. The system will be unable to distinguish between actual drive instability and instantaneous mechanical resonance, resulting in frequent control oscillations or unnecessary power interruptions.

[0038] To distinguish between these two, this step monitors the time-domain evolution characteristics of the electromechanical energy decoupling index, combines it with the frequency-domain response of wheel kinematics, and uses cross-correlation analysis within a sliding window to obtain the road surface change confidence index. This ensures that the system only triggers a response when the actual road surface condition changes, avoiding unnecessary control actions. The road surface change confidence index is calculated as follows: ; In the formula, C is the pavement abrupt change confidence index; it represents the confidence level that the currently observed dynamic anomaly is caused by a sudden change in the actual pavement adhesion performance, rather than by random mechanical impact; its value range is... When C is close to 0, the system determines it as normal fluctuation or external disturbance and does not trigger compensation; when C is close to 1, the system is highly certain that road surface adhesion has deteriorated and immediately activates transient torque limiting. γ is a road surface feature normalization factor used to precisely balance the system's sensitivity to road surface changes and its ability to prevent accidental triggering. When the value of γ is too small, the rate of increase of the road surface change confidence index C is significantly reduced, resulting in a delay in the response of the torque limiting command and an inability to suppress severe slippage in time. When the value of γ is too large, the system becomes overly sensitive to minor vibrations or sensor noise, frequently misjudging them as road surface changes, causing unnecessary power interruptions and reducing operational continuity. Therefore, the reasonable range of γ is strictly limited to [0.4, 0.6]. In this embodiment, γ = 0.5 is used, which achieves the optimal balance between safety constraints and smooth operation. In practical applications, the implementer can dynamically adjust γ according to the vehicle's operating conditions: for special engineering vehicles where safety is paramount, γ can be increased to 0.6 to enhance the response to sudden changes; for logistics and transfer vehicles with efficiency orientation, γ can be decreased to 0.4 to suppress accidental triggering and ensure operational continuity. The sliding time window length represents the depth of historical data analysis. If L is too small, the data within the window is easily affected by single-point noise, causing the C value to fluctuate and triggering erroneous actions. If L is too large, it is impossible to capture the initial characteristics of millisecond-level road surface changes, thus losing the advantage of blind spot pre-intervention. In this embodiment, a window depth of 10ms is set, which precisely covers the initial dynamic process of a typical wheel passing through a small obstacle or entering a low-adhesion area, effectively smoothing high-frequency interference while ensuring rapid response. In other embodiments, the implementer can fine-tune the window depth within the [8,12] period range according to the vehicle speed or sensor signal-to-noise ratio. The rate of change of the electromechanical energy decoupling index at the i-th sampling point in the sliding window represents the burst speed of energy instability. is the transient impact characteristic of the electromechanical energy decoupling index at the i-th sampling point of the sliding window, is the derivative of the wheel angular acceleration, and represents the transient mechanical impact intensity experienced by the wheel.

[0039] This calculation formula generates a normalized road surface abrupt change confidence index by multiplying and summing the rate of change of the electromechanical energy decoupling index with the derivative of the wheel angular acceleration point by point within a short-time sliding window, and then performing an exponential mapping. This index is used to determine whether a true abrupt change has occurred in road surface adhesion. When entering a low-traction surface, the electromechanical energy decoupling index rises rapidly, and at the same time, the wheels lose restraint, causing a sudden surge in speed and generating an impact; at this moment, the product of these two factors increases dramatically, leading to an exponential increase. Approaching This causes the road surface change confidence index to rapidly increase. Upon approaching, the system determined that a highly plausible road surface abrupt change had occurred.

[0040] When a vehicle runs over a rock, although a significant mechanical impact occurs due to the instantaneous load change, the energy output by the motor can still be effectively converted into vehicle kinetic energy because the road surface medium remains unchanged. The rate of change of the electromechanical energy decoupling index remains at a low level. Since energy and motion are not substantially decoupled, the product term is small, and the road surface change confidence index will remain close to [a certain level]. If the value is low, the system will not trigger compensation.

[0041] S4: Obtain transient corrected boundary based on road surface change confidence index combined with slip energy.

[0042] It should be noted that in the initial stage of a step drop in road surface adhesion performance, the conventional identification algorithm within the system is limited by the sampling window and computation convergence time. Its output identification value often remains at the old high adhesion level and cannot truly reflect the current physical limit. If this lagging identification value is directly used as the hard constraint for model predictive control, it will cause the execution end to issue torque commands that exceed the actual bearing capacity of the ground, resulting in irreversible and severe slippage. In order to establish an effective protection boundary in the blind zone before the identification algorithm converges, this step introduces a transient correction boundary to provide real-time hard constraints for model predictive control.

[0043] Wherein, the transient correction boundary Combining static parameter identification, transient slip intensity, and time accumulation effect, the formula for calculating the transient correction boundary is: ; In the formula, The transient correction boundary serves as the upper limit of real-time constraints for the model predictive control solver. The conventional identification value is estimated by recursive least squares method. Due to the influence of the algorithm's convergence speed, it has a significant lag at the moment of abrupt change; C is the road surface change confidence index. The real-time slip energy at the current moment is defined as follows: , This refers to the electromagnetic torque of the motor. The wheel's angular velocity; For effective ground traction; The longitudinal speed of the vehicle; representing the power loss during the freewheeling of the drive wheels; The preset maximum slip energy threshold is calibrated based on tire wear limit and operating conditions. This is the attenuation gain coefficient, in units of 1 / s, used to adjust the suppression strength of the time accumulation effect of the road surface change confidence index on the transient correction boundary. If the value of β is set too low, the system will be insensitive to continuous slight slippage, and the transient correction boundary will not decrease sufficiently, which may lead to prolonged tire slippage and temperature accumulation, increasing the risk of wear and thermal failure. If the value of β is set too high, it will overreact to short-term reasonable slippage such as starting or overcoming obstacles, causing the transient correction boundary to be unnecessarily lowered, resulting in conservative torque limiting and weakening traction performance. In this embodiment, β is set to 0.25, which effectively suppresses energy waste and tire damage. In other embodiments, implementers can fine-tune the value within the period range of [0.15, 0.35] according to task requirements. The upper limit should be used in scenarios where durability is prioritized, and the lower limit should be used in scenarios where mobility is prioritized. This is a pavement anomaly persistence factor; it is obtained by time-domain accumulation or integration of the pavement abrupt change confidence index within the sliding time window. Reflecting the sustained intensity of recent abnormal road surface conditions, when When it remains at a high level for an extended period Gradually increase, thereby enhancing the suppression effect on the transient correction boundary; if Only instantaneous pulses, then Rapid decay, avoid being overly conservative.

[0044] This step achieves the following technical effect through a dual dynamic constraint mechanism of proportional and integral constraints: (1) Proportional instantaneous suppression mechanism: When a vehicle switches from a high-adhesion road surface to a low-adhesion road surface at the instant, the road surface change confidence index rapidly approaches 1; at this time, due to the limitation of the algorithm window, the conventional identification value It remains at a high level; with real-time slip energy The surge, the proportional correction term in the calculation formula It rapidly reduces the rate of change; this proportional instantaneous suppression mechanism can overcome the convergence delay of the parameter identification algorithm and approach the real physical limit in a very short time, achieving real-time compensation for the underflow of the adhesion coefficient.

[0045] (2) Integral Persistent Constraint Mechanism: If the road surface adhesion deterioration continues, causing the road surface abrupt change confidence index to remain high over multiple sampling periods, the integral term in the calculation formula will be constrained. This will generate a continuously increasing negative bias; the integral persistence constraint mechanism further dynamically lowers the adhesion upper limit constraint during the ongoing process where the slippage state has not been eliminated. This offsets the torque overshoot that may be caused by the mechanical inertia of the hub motor and the measurement error of the sensor, prevents the motor from continuously outputting invalid torque in a low-adhesion environment, and ensures that the wheel end speed is always in a stable and controlled range; when the calculated value is lower than 0, the transient correction boundary is automatically set to 0 to avoid the failure of physical constraints.

[0046] S5: Substitute the transient correction boundary into the cost function of adaptive model predictive control to solve for the optimal electromagnetic torque sequence that satisfies the ground adhesion limit. Send the first term as an instruction to the driver to control the motor to output the torque, thereby achieving adaptive energy efficiency control.

[0047] It should be noted that in complex closed-loop control processes, the system faces a state of ambiguity due to a mismatch between perception and execution. That is, if the controller's cost function only focuses on the vehicle's power response or energy consumption reduction, without a mandatory perception of the ground adhesion limit, then there will be a physical envelope gap between the calculated torque command and the torque that the ground can actually bear. At the moment of sudden change in road surface, this disconnect will cause the solver to issue a logically optimal but physically excessive torque, triggering secondary slippage or control oscillation.

[0048] To eliminate the risk of physical runaway at the execution end and achieve a closed loop between perception and control, this step uses the transient correction boundary as a time-varying hard constraint condition, which is solved in conjunction with the cost function. The cost function balances the slip ratio tracking accuracy and motor energy consumption, ensuring that the drive wheels operate within the optimal slip range. The constraint condition forcibly limits the torque within the ground adhesion capability range. By solving the constrained optimization problem, the system generates the optimal torque sequence that meets safety requirements, achieving closed-loop control from perception to execution and ensuring that control commands always conform to physical feasibility.

[0049] Within each control cycle, the system constructs a cost function in the defined prediction time domain; the calculation formula is as follows: ; In the formula, The cost function to be minimized reflects the comprehensive evaluation of control performance and energy consumption over the entire prediction time domain; N is the prediction time domain length in model predictive control, that is, the controller predicts the system behavior over the next N control cycles at the current moment and optimizes the control sequence within this range. To predict the first in the time domain The real-time slip ratio of the step is equal to the actual vehicle speed minus the wheel circumferential speed divided by the actual vehicle speed, which represents the current grip state of the drive wheel. The target slip ratio is usually set at the optimal slip ratio corresponding to the peak value of the tire adhesion coefficient. Let K be the electromagnetic torque sequence of the k-th motor to be solved. The equivalent internal resistance coefficient of the motor; and For the preset weighting coefficients, satisfy Since slip ratio tracking accuracy directly determines the effectiveness of vehicle traction and driving stability, while energy consumption optimization only affects energy efficiency, slip ratio tracking is given higher priority, meaning it requires... In this embodiment, the following settings are provided: , In other embodiments, implementers can adjust the weighting coefficients within a reasonable range according to actual working conditions. and For scenarios with extremely high requirements for traction reliability, it can be Increased to [0.7, 0.8]; for long-term operation scenarios sensitive to energy efficiency and temperature rise, the [value] can be increased to [0.7, 0.8]. Increase to [0.4, 0.5]; This is a preset reference torque, in N·m. This is the preset reference internal resistance coefficient, in Ω or normalized impedance.

[0050] This cost function achieves energy efficiency optimization while ensuring vehicle traction performance by weighted balancing two objectives: slip ratio tracking accuracy and motor energy consumption. The calculation uses the squared slip ratio deviation at each step in the prediction time domain to measure control accuracy, ensuring the drive wheels operate within the optimal slip range where tire adhesion is strongest. Assess the Joule heat loss of the motor and suppress unnecessary high torque output.

[0051] To ensure that the torque sequence obtained from the cost function does not exceed the physical limits of the ground, this scheme uses the transient corrected boundary obtained in the preceding steps as a time-varying hard constraint condition for the quadratic programming problem. This means that during the optimization process of the cost function, the electromagnetic torque sequence of the motor to be solved must strictly satisfy the following inequality: ; In the formula, To predict the optimal electromagnetic torque sequence of the motor to be solved in the time domain; For transient boundary correction; The real-time vertical load at the wheel end is measured by the sensor; This is the wheel's rolling radius.

[0052] The system solves the constrained quadratic programming problem online, and then incorporates the transient correction boundary into the optimization framework of adaptive model predictive control as a hard constraint condition and cost function to generate the optimal electromagnetic torque sequence that satisfies the ground physical adhesion limit. Based on the rolling optimization principle of MPC, only the first element of this sequence is sent to the hub motor driver as the execution command for the current sampling period, ensuring that the motor output torque is always limited within the actual ground adhesion capability when the road adhesion coefficient undergoes a step change. At the next sampling moment, the system recalculates the state signal based on updated multi-source sensor data and executes the above complete optimization process again, guiding the new energy vehicle to automatically optimize torque output when the road adhesion changes, realizing the closed-loop execution of energy efficiency improvement and anti-skid control.

[0053] For example, Figure 2 The diagram shows a comparison of wheel-end linear velocity responses, illustrating the dynamic performance of two control strategies under conditions of a sudden change in road surface adhesion. After the sudden change, the traditional scheme, based on historical data identification, fails to detect the sudden drop in ground bearing capacity due to parameter convergence lag, resulting in a rapid increase in drive wheel speed and a significant deviation from the desired driving speed, forming a clear uncontrolled spinning range. In contrast, the adaptive scheme proposed in this invention relies on the millisecond-level perception capability of electromechanical energy decoupling indicators for changes in road surface conditions, instantly generating transient constraint boundaries that match the current adhesion limit. This effectively suppresses abnormal wheel speed increases, allowing the vehicle's motion state to quickly return to stability, fully demonstrating its strong robustness and responsiveness in sudden low-adhesion scenarios.

[0054] For example, Figure 3 The diagram shows a comparison of motor output torque commands. Traditional methods maintain high torque commands even after road surface deterioration, significantly exceeding the upper limit of friction that the ground can actually provide. In contrast, this invention actively reduces torque output at the moment of sudden change and dynamically adjusts it according to the sustained intensity of road surface anomalies, ensuring that the drive command is always within the physically feasible range, thereby avoiding slippage or traction collapse caused by overload.

[0055] For example, Figure 4 The diagram shows the cumulative ineffective energy loss. Traditional control cannot identify attachment drops during the critical window period, and a large amount of electrical energy is continuously converted into useless sliding kinetic energy, resulting in energy waste accumulating rapidly over time. In contrast, this invention significantly reduces the ineffective slippage of the drive wheels by tightening the upper limit of torque in real time, keeping the accumulated energy consumption at a low level and greatly reducing the total energy expenditure.

Claims

1. An adaptive optimization method for energy efficiency control of hub motors in new energy vehicles, characterized in that, include: The system acquires the operating status data of the hub motors of new energy vehicles at current and historical times, and performs noise reduction processing on the operating status data to obtain high-confidence status signals. The real-time matching relationship between the motor input power and the rate of change of vehicle mechanical kinetic energy is evaluated based on the state signal to obtain the electromechanical energy decoupling index. Based on the time-domain evolution characteristics of the electromechanical energy decoupling index and combined with the frequency-domain response of wheel kinematics, the road surface mutation confidence index is calculated to evaluate the confidence level that the dynamic anomaly is caused by the mutation of the actual road surface adhesion performance. Based on the road surface mutation confidence index, combined with the real-time slip energy lost in the drive wheel idling at the current moment, a transient correction boundary is obtained to limit the upper limit of the model predictive control solution. The transient correction boundary is substituted as a time-varying hard constraint into the cost function of the adaptive model predictive control to solve for the optimal electromagnetic torque sequence that satisfies the ground adhesion limit. The first term of the optimal electromagnetic torque sequence is sent as an instruction to the hub motor driver, and the driver uses closed-loop control to output the torque of the motor, thereby realizing adaptive energy efficiency control.

2. The adaptive optimization energy efficiency control method for hub motors in new energy vehicles according to claim 1, characterized in that, The operating status data includes motor electromagnetic torque, real-time wheel angular velocity, vehicle longitudinal acceleration, wheel end vertical load, and vehicle speed sensing signal; real-time wheel angular acceleration and vehicle longitudinal centerline velocity; and the equivalent translational mass of a single wheel, the moment of inertia of the wheel and its transmission components, the wheel rolling radius, and the equivalent internal resistance coefficient of the motor, all pre-stored in the control unit.

3. The adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to claim 2, characterized in that, The formula for calculating the electromechanical energy decoupling index is: ； In the formula, For electromechanical energy decoupling indicators; This represents the current real-time electromagnetic input power of the motor. The equivalent translational mass of a single wheel; The speed along the longitudinal centerline of the vehicle; The longitudinal acceleration of the vehicle; The moment of inertia of the wheel and its transmission components; This refers to the real-time angular velocity of the wheel. This refers to the real-time angular acceleration of the wheel; This represents the average input power of the motor at the current moment. It is a very small constant.

4. The adaptive optimization energy efficiency control method for hub motors in new energy vehicles according to claim 1, characterized in that, The formula for calculating the road surface abrupt change confidence index is: ； In the formula, The confidence index for road surface abrupt changes; This is a road surface feature normalization factor. The length of the sliding time window; The rate of change of the electromechanical energy decoupling index at the i-th sampling point in the sliding time window; This is the transient impact characteristic of the electromechanical energy decoupling index at the i-th sampling point in the sliding window.

5. The adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to claim 1, characterized in that, The formula for calculating the transient correction boundary is: ； In the formula, For transient boundary correction; The standard identification value estimated by recursive least squares method; C represents the confidence index for road surface abrupt change; The real-time slip energy at the current moment; The maximum allowable slip energy threshold; This is the attenuation gain coefficient. This is a factor contributing to the persistence of road surface anomalies.

6. The adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to claim 5, characterized in that, The real-time slip energy at the current moment is equal to the product of the motor's electromagnetic torque and the wheel's angular velocity minus the product of the effective ground traction force and the vehicle's longitudinal velocity.

7. The adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to claim 1, characterized in that, The cost function is calculated as follows: ； In the formula, Let be the cost function to be minimized; N is the prediction time domain length in model predictive control. To predict the first in the time domain Real-time slip rate of the step; The target slip ratio; Let K be the electromagnetic torque sequence of the k-th motor to be solved. The equivalent internal resistance coefficient of the motor; and Preset weighting coefficients; The preset reference torque; This is the preset reference internal resistance coefficient.

8. The adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to claim 2, characterized in that, The constraints are as follows: ； In the formula, To predict the optimal electromagnetic torque sequence of the motor to be solved in the time domain; Correct the boundary for transient attachment; The real-time vertical load at the wheel end is measured by the sensor; This is the rolling radius of the wheel.

9. The adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to claim 7, characterized in that, The preset weighting coefficient satisfies And require .

10. The adaptive optimization method for energy efficiency control of hub motors in new energy vehicles according to claim 1, characterized in that, The noise reduction method mentioned refers to Kalman filtering.

Images