Commercial vehicle dual motor system based on complementary high-efficiency region and torque distribution control method

Abstract

The application provides a commercial vehicle double-motor system and a torque distribution control method based on complementary high-efficiency zones, and solves the problems existing in the torque distribution technology of the current double-motor drive axle vehicle.The double-motor system comprises a torque axle and a speed axle, the torque axle is configured with a high-torque-density motor and a large-reduction-ratio reducer, and the high-efficiency zone is concentrated in a low-rotational-speed and high-torque region; the speed axle is configured with a high-power-density motor and a small-reduction-ratio reducer, and the high-efficiency zone is concentrated in a high-rotational-speed and medium-low-torque region, the high-efficiency zones of the two axles are complementary on a rotational speed-torque plane, and jointly cover the full-working-condition operation region of the vehicle; when a large torque is needed, the torque axle shares a large output power; when a high vehicle speed is needed, the speed axle shares a large output power, and the output power of the high-efficiency zone is always more. The application forms complementary high-efficiency zones through the differential design of the double axles, realizes the lowest comprehensive energy consumption of the vehicle in the full-working-condition range, and simultaneously considers the power performance and the drivability.

Description

Technical Field

[0001] This invention belongs to the field of electric drive axle commercial vehicle drive motor control technology, specifically involving a torque distribution control method and system that achieves optimal energy consumption under all operating conditions by differentiating the efficiency characteristics of dual electric drive axles and integrating multiple parameters such as vehicle speed, throttle opening, and vehicle acceleration. Background Technology

[0002] The current torque distribution technology for dual-electric drive axle vehicles mainly faces the following technical bottlenecks:

[0003] 1) The design of motors with similar characteristics leads to overlapping high-efficiency zones.

[0004] In existing technologies, the front and rear drive axles mostly use motors of the same specifications, and their efficiency MAPs (multi-dimensional parameters such as motor torque, speed, and temperature, corresponding to a limit efficiency) highly overlap, making it impossible to form an effective complementarity. Throughout the entire operating range of the vehicle (especially during low-speed, high-torque climbing and high-speed cruising), the application of the inefficient operating range of the dual motors cannot be minimized, resulting in higher overall energy consumption.

[0005] 2) Insufficient intelligence in allocation strategies

[0006] Existing torque distribution algorithms mostly employ fixed ratios or simple rules based on a single parameter, such as switching distribution modes solely based on vehicle speed. These methods fail to fully integrate key dynamic parameters such as throttle opening (reflecting the driver's real-time intentions) and vehicle acceleration (reflecting actual dynamic load), resulting in a low degree of matching between the distribution strategy and real-time operating conditions.

[0007] 3) The algorithm has a single optimization objective and lacks a global perspective.

[0008] Most control strategies only aim to minimize instantaneous energy consumption and lack global optimization capabilities based on operating condition prediction. Furthermore, the algorithms' utilization of the motor efficiency MAP is rather crude, failing to leverage the synergistic advantages of complementary high-efficiency zones.

[0009] 4) The coupling with the vehicle's longitudinal dynamics is not tight.

[0010] Existing solutions do not deeply couple torque distribution with the vehicle's longitudinal dynamics model (including inertial forces, driving resistance, etc.), and therefore cannot achieve the optimal balance between power response and energy efficiency under transient conditions such as rapid acceleration and hill climbing. Summary of the Invention

[0011] This invention aims to provide an innovative torque distribution control method and system for dual-electric drive axle vehicles. By differentiating the design of the two axles to form complementary high-efficiency zones, and developing an intelligent distribution algorithm strongly coupled with vehicle speed, throttle opening, and vehicle acceleration, the invention achieves the lowest comprehensive energy consumption of the vehicle across all operating conditions, while also taking into account power and drivability.

[0012] This invention is achieved through the following technical solutions:

[0013] A dual-motor system for commercial vehicles based on complementary high-efficiency zones includes a torque bridge and a speed bridge. The torque bridge is equipped with a high torque density motor and a large reduction ratio reducer, with its high-efficiency zone concentrated in the low-speed, high-torque region, where low speed and high torque refer to 0%-60% of the rated speed and 30%-100% of the peak torque. The speed bridge is equipped with a high power density motor and a small reduction ratio reducer, with its high-efficiency zone concentrated in the high-speed, low-to-medium torque region, where high speed and low-to-medium torque refer to 30%-100% of the rated speed and 10%-60% of the peak torque. The high-efficiency zones of the two bridges complement each other on the speed-torque plane, jointly covering the entire operating range of the vehicle. When high torque is required, the torque bridge shares a larger share of the output power. When high vehicle speed is required, the speed bridge shares a larger share of the output power, always maintaining a higher proportion of output power in the high-efficiency zone.

[0014] A torque distribution control method for a dual-motor system in a commercial vehicle based on complementary high-efficiency zones includes the following steps:

[0015] The first step is multi-source signal acquisition and fusion processing;

[0016] The second step is to calculate the total torque requirement for the entire vehicle.

[0017] The third step is the intelligent working condition recognition algorithm;

[0018] The fourth step is the multi-mode adaptive torque distribution algorithm.

[0019] Step 5: Adaptive learning.

[0020] This invention proposes and implements a dynamic torque distribution intelligent decision-making mechanism based on the complementary efficiency characteristics of dual motors, aiming at achieving "real-time optimal global system efficiency." It elevates drive torque distribution from a relatively static, rule-based control problem to a dynamic, optimization-model-based system-level energy efficiency management problem, thus realizing the optimal control method. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of a dual-motor system for a 6×4 commercial vehicle.

[0022] Figure 2 This is a schematic diagram of the complementary efficiency map of dual motors in a 6×4 commercial vehicle.

[0023] Figure 3 This is a schematic diagram illustrating the calculation process for the total torque demand of the entire vehicle.

[0024] Figure 4 The torque required by the driver Calculation flowchart;

[0025] Figure 5 It is the driving resistance torque. Calculation flowchart;

[0026] Figure 6 This is a membership diagram of the set of vehicle speeds;

[0027] Figure 7 This is a flowchart of the multi-mode adaptive torque distribution algorithm. Detailed Implementation

[0028] The present invention relates to a torque distribution control method and system for a dual-motor system in a commercial vehicle based on complementary high-efficiency zones, comprising two parts: innovative design of a differentiated dual-electric drive axle architecture and innovative control algorithm.

[0029] 1. Differentiated dual-electric drive bridge architecture

[0030] like Figure 1 , 2 As shown, this invention proposes to design the dual electric drive bridge as two functionally complementary subsystems:

[0031] 1) Torque bridge

[0032] It is equipped with a high torque density motor and a large reduction ratio reducer, which are specially optimized for low-speed, high-torque conditions. Its high-efficiency range is concentrated in the low-speed, high-torque range (such as 0%-60% of rated speed and 30%-100% of peak torque, which can be determined after vehicle calibration).

[0033] 2) Speed ​​bridge

[0034] It is equipped with a high-power-density motor and a low-reduction-ratio reducer, which are specifically optimized for high-speed cruising efficiency. Its high-efficiency range is concentrated in the high-speed and low-to-medium torque range (such as 30%-100% of the rated speed, or 10%-60% of the peak torque, which is determined after vehicle calibration).

[0035] The high-efficiency zones of the two axles complement each other on the speed-torque plane, jointly covering the entire operating range of the vehicle. Their complementary relationship is as follows: Figure 2 As shown.

[0036] The two sets of bridges are a speed bridge and a torque bridge. The high-efficiency range of the motor complements each other under different vehicle speeds and loads. When high torque is required, the torque bridge takes on more output power. When high vehicle speed is required, the speed bridge takes on more output power, always maintaining more output power in the high-efficiency range to achieve the goal of energy saving.

[0037] 2. Control Algorithm

[0038] Power allocation is correlated with vehicle speed, throttle opening, and vehicle acceleration, with optimal system efficiency as the constraint to minimize vehicle energy consumption. Specifically, this includes the following steps:

[0039] 2.1 First step: Multi-source signal acquisition and fusion processing

[0040] The overall vehicle operating status and the status of component systems are transmitted to the vehicle control unit (VCU) input interface via various sensors or the component system's own controller (BMS, TCU, etc.). The main signals transmitted include:

[0041] Vehicle speed v (km / h);

[0042] Switch opening (%, normalized to 0-1);

[0043] Vehicle longitudinal acceleration α (m / s 2 ), obtained through estimation by the vehicle control unit (VCU);

[0044] Motor speed ( State parameters such as temperature (T) and battery SOC (%);

[0045] As a state vector, the original signal is denoised and fused using a Kalman filter algorithm called in the application layer software of the vehicle control unit (VCU). This further reduces noise and improves signal quality, thereby enhancing the robustness and accuracy of the control system. The five equations of the Kalman filter algorithm are as follows:

[0046] State prediction equation:

[0047] Error covariance prediction equation:

[0048]

[0049] Kalman gain prediction equation:

[0050]

[0051] State update equation:

[0052]

[0053] Error covariance update equation:

[0054]

[0055] in, For the state vector ([ Z is the observation vector, F and H are the state transition and observation matrices, P is the estimation error covariance matrix, and Q and R are the process noise and measurement noise covariance matrices. This is the Kalman gain.

[0056] k is the time step, k|k means that information up to and including the latest observation data at time k is used at time k. k|k-1 means that information at time k-1 is given at time k-1. k-1|k-1 means that information at time k-1 has been fused with the latest observation data at time k-1.

[0057] 2.2 Second step, calculation of total torque demand for the whole vehicle

[0058] Only by knowing "how much torque the whole vehicle needs in total" can we optimize "how to distribute that torque" and the total required torque. The calculations integrate driver intent and vehicle dynamics requirements:

[0059]

[0060] Vehicle torque requirements

[0061] This indicates the torque demanded by the driver, derived from the accelerator pedal opening. It reflects the driver's intentions, and responsiveness can be emphasized during allocation;

[0062] It reflects dynamic changes, and smoothness can be emphasized when allocating resources;

[0063] This represents the driving resistance torque, which originates from air resistance, rolling resistance, and gradient resistance. It reflects the steady-state load and can be distributed with an emphasis on efficiency.

[0064] The calculation process for the total torque requirement of the whole vehicle can be found here. Figure 3 .

[0065] The calculations for each component are as follows:

[0066] 2.2.1 Driver's required torque :

[0067]

[0068] In the formula, The basic torque table is a static mapping based on a two-dimensional lookup table, defining the basic torque ( ) and switch opening ( ), vehicle speed ( The relationship between ) is that each ( The coordinates correspond to a basic torque. The values ​​in this table are determined through vehicle calibration and placed in the program of the vehicle control unit (VCU). They are retrieved and called in real time during calculation. , This is the proportional-derivative coefficient, used to improve responsiveness. Driver-required torque. See the calculation process. Figure 4 .

[0069] Meaning of each parameter:

[0070] 1) Dynamic compensation item The system compensates for changes in throttle opening to capture the driver's transient intentions. Specifically:

[0071] The difference between the original switch opening and the filtered opening.

[0072] When the driver quickly presses the pedal: The compensation term is positive, which increases torque output and improves response speed;

[0073] When the driver quickly releases the pedal: The compensation term is negative, which rapidly reduces torque and enhances energy recovery.

[0074] 2) Parameters Proportional gain, typical value 0.1-0.5 (can be adjusted according to specific operating conditions). When calibrating, both responsiveness and smoothness should be considered.

[0075] 3) Differential compensation term This indicates compensation for the rate of change of throttle opening, which helps predict the driver's intentions. The rate of change of the switch opening (% / s):

[0076] When the driver slams down the pedal: It is large and positive, the compensation term increases torque and provides acceleration;

[0077] As the driver slowly depresses the pedal: The compensation is relatively small, ensuring smooth acceleration.

[0078] parameter : Differential gain, typical value 0.05-0.2 (can be calibrated and adjusted according to specific operating conditions).

[0079] 2.2.2 Inertia Compensation Torque :

[0080]

[0081]

[0082] In the formula, For the equivalent moment of inertia, For the tire radius, The overall speed ratio, For motor inertia, For the overall vehicle weight.

[0083] 2.2.3 Driving resistance torque :

[0084]

[0085]

[0086]

[0087]

[0088] In the formula, For air resistance, air density, Where A is the drag coefficient and A is the effective frontal area of ​​the cab. The speed is the vehicle speed.

[0089] For rolling resistance, For the total vehicle weight, It is the acceleration due to gravity. θ is the rolling resistance coefficient, and θ is the road slope (measured by the slope sensor integrated in the transmission controller TCU).

[0090] This is slope resistance.

[0091] Driving resistance torque See the calculation process. Figure 5 .

[0092] These three formulas transform the vague "driver's pedal input" action into precise physical torque requirements, forming the premise and foundation for the intelligent, efficient, and accurate torque distribution achieved in this patent. These formulas precisely calculate the total required torque. This forms the basis for subsequent torque distribution optimization.

[0093] The engineering implementation of the calculation requires three steps:

[0094] 1) Parameter calibration process

[0095] Phase 1: Basic Calibration

[0096] a) Static mapping Calibration: Record the differences on the drum test bench or road. The base torque under the combination of ,v)

[0097] b) Rolling resistance coefficient Calibration: Fit the gliding resistance curve through gliding tests;

[0098] c) Drag coefficient Calibration: Wind tunnel test or high-speed taxiing test.

[0099] Phase 2: Dynamic Parameter Calibration

[0100] a) , Calibration: This can be performed at the test track using Tip-in / Tip-out (rapidly pressing / releasing the accelerator) tests, creep start tests, full-throttle acceleration tests, and coasting tests until the optimal response speed and smoothness are achieved, thus determining the final calibration. , ;

[0101] b) Filter parameter calibration: Select appropriate filter coefficients based on the noise characteristics of the pedal sensor.

[0102] Phase 3: Adaptive Learning

[0103] Vehicle quality Learning: Accelerate data through multiple starts and use it to infer quality.

[0104] 2) Real-time computing optimization

[0105] a) Under the constraints of limited storage space and limited VCU computing power, in order to obtain continuous, smooth, and accurate torque output and avoid step changes in motor output torque due to sparse lookup points, static mapping needs to be optimized. .

[0106] For example, the β axis has 20 data points. When the actual input When the value falls between two calibration points, taking the value of the closest point will result in discontinuous output torque, affecting driving smoothness.

[0107] b) Regarding the rate of change of the switch opening signal The differential method with filtering is used to avoid noise amplification, which could cause high-frequency jitter in the torque command. In engineering implementation, the VCU first performs low-pass filtering and then differential calculation to obtain a smooth and reliable switch opening signal.

[0108] c) Calculate air resistance When the vehicle speed v is a square term, in order to reduce the number of multiplication operations and lower the computational load on the VCU, a recursive calculation method needs to be used in the VCU program to reduce the computational load.

[0109] 3) Special working condition handling

[0110] a) Coasting / Brake: When When the torque value is close to zero or negative, the vehicle controller (VCU) sends a negative torque command to the motor controller (MCU), causing the motor to operate in reverse drag mode and initiating energy recovery.

[0111] b) Extreme operating conditions (such as peak power exceeding limits, low-friction road driving, etc.): When the calculated... When the system exceeds its capacity, the torque is continuously reduced according to a pre-defined strategy (usually following the typical priority order of safety > basic regulatory functions > performance > comfort) to adapt to the current operating conditions.

[0112] c) Sensor failure: The vehicle control unit (VCU) performs gradient calculations based on wheel speed and motor torque.

[0113] 2.3 The third step, intelligent working condition recognition algorithm

[0114] Intelligent operating condition recognition algorithm:

[0115] Based on vehicle speed v, throttle opening ß, and vehicle acceleration α, a method combining fuzzy logic and rule-based judgment is used to identify operating conditions. The corresponding torque logic is as follows: Figure 3 As shown.

[0116] 1) Define the fuzzy set of input variables:

[0117] a) Vehicle speed v: {very low, low, medium, high, very high};

[0118] b) Switch opening ß: {Very small, small, medium, large, very large};

[0119] c) Acceleration Demand Index : {very small, small, medium, large, very large};

[0120] See the membership diagram of the vehicle speed set. Figure 6 It can be adjusted according to the actual working conditions.

[0121] Example calculation

[0122] The membership function expression for a triangle is: a, b, and c are the left boundary, peak point, and right boundary of the triangle, respectively.

[0123] When v=25km / h, it is located in the overlapping segment of the right falling edge of the "very low" set and the left rising edge of the "low" set.

[0124] Belongs to the category of "very low" membership degree: ;

[0125] Membership degree that falls under the category of "low": ;

[0126] Membership degree to other sets: 0.

[0127] 2) Based on vehicle dynamics theory, motor efficiency MAP characteristics, whole vehicle simulation analysis results, and real vehicle calibration test data, a fuzzy rule base is comprehensively formulated, which includes several rules corresponding to several working conditions.

[0128] Simultaneously, based on statistical analysis of historical data and energy consumption calculations based on simulation optimization, the confidence level of each rule is determined. The confidence level characterizes the determinism of the rule in actual vehicle operation; when multiple rules conflict, the rule with higher confidence level has a higher weight.

[0129] Example: In a certain region, for a 49-ton tractor truck, the following rules are determined based on calculations and calibration:

[0130] Rule 1: IF v IS is very low AND IS is very high in the operating condition, dominated by high torque (confidence level 0.9). Efficiency strategy: Torque bridge as the primary factor;

[0131] Rule 2: IF v IS very high AND IS is very small. THEN operating condition IS is dominated by high speed (confidence level 0.85). Efficiency strategy: speed axle is the primary factor;

[0132] Rule 3: IF > 0.6 THEN operating condition IS transient acceleration (confidence level 0.8). Efficiency strategy: dual-axle cooperation, emphasizing the torque axle;

[0133] Rule 4: IF v IS AND α IS THEN IS combined operating condition (confidence level 0.7). Efficiency strategy: Global optimization;

[0134] Rule 5: IF v IS low AND α IS small AND Iacc > 0.5 THEN Operating condition IS gentle uphill (confidence level 0.75). Efficiency strategy: comprehensive allocation;

[0135] Rule 6: IF v IS High AND Iacc < -0.3 THEN Operating condition IS Downhill deceleration (confidence level 0.8). Efficiency strategy: Energy recovery.

[0136] 3) Finally, the weighted average method is used for defuzzification, and the working condition type and confidence level are output:

[0137]

[0138] a) The explanation of mode i is as follows:

[0139] `mode i` is the numerical label corresponding to each fuzzy rule, similar to an "ID number," mapping different operating conditions to specific positions within the continuous interval [0,1]. This value is preset and reflects the degree of torque distribution tendency, as shown in Table 1:

[0140] Table 1. Examples of mode_i settings in the rule base.

[0141]

[0142] Value setting principles:

[0143] Orderliness: The operating conditions are arranged in the order from the speed bridge to the torque bridge;

[0144] Continuity, adjacent operating conditions Values ​​are close, resulting in a smooth transition;

[0145] Non-overlapping, different working conditions The values ​​have reasonable intervals;

[0146] In terms of physical meaning, the magnitude of the value reflects the degree of torque distribution tendency.

[0147] The setting process requires combining simulation analysis and actual vehicle calibration adjustments to ensure that the errors between the setpoints and calculated values ​​meet control requirements (e.g., error ≤ 10%). This is an iterative process. Ultimately, in the vehicle control unit (VCU), Values ​​are stored in the form of lookup tables or constants.

[0148] b) The explanation is as follows:

[0149] It is the rule activation weight, which represents the minimum membership value among the current input variables (such as vehicle speed, throttle opening, etc.) that triggers the rule.

[0150] For example:

[0151] Rule 1: IF v IS is very low AND IS is very large THEN. Under operating conditions, IS is dominated by high torque.

[0152] Suppose that the current vehicle speed v belongs to the fuzzy set with a membership degree of "very low" = 0.8;

[0153] Current throttle opening The membership degree of the fuzzy set "very large" is 0.4;

[0154] Vehicle acceleration = 0.65, membership degree = 1 (Rule 3 is a binary condition, the activation weight is either 1 or 0).

[0155] but =min(0.5, 0.4)=0.4, confidence level=0.9;

[0156] =1.0, confidence level =0.8;

[0157] c) The explanation is as follows:

[0158] It is the defuzzified output value, which is the working condition type determined by the algorithm.

[0159] Example:

[0160] In this invention, for a certain 6×4 vehicle model For example, the value is set. The value is determined as follows:

[0161] when When high vehicle speed is the dominant operating condition, the optimal allocation strategy for speed bridge efficiency should be adopted first.

[0162] when Under comprehensive operating conditions, a model-based global optimization algorithm is employed, in which... The closer the value is to 0.7, the higher the dynamic weight in the algorithm;

[0163] when During transient acceleration, the torque axle takes priority.

[0164] when At high torque conditions, the torque bridge dominates.

[0165] above The threshold range is determined based on the following criteria:

[0166] First, based on the complementary characteristics of the dual-motor efficiency MAP, numerical analysis is performed on the efficiency contour plots of the torque bridge and speed bridge. On the speed-torque two-dimensional plane, contour lines (η) representing the efficiency difference between the two bridges are plotted. 扭矩桥 - η 速度桥 When the efficiency difference is consistently positive and greater than a certain threshold (e.g., 5%), it indicates that the torque bridge has a significant efficiency advantage, and the corresponding Mode should be biased towards 1.0; otherwise, it should be biased towards 0.0.

[0167] The basis for the threshold of 0.7: Simulation data shows that in the comprehensive working condition range (Mode≈0.5-0.7), the efficiency of the two bridges is comparable, and the best effect can be obtained by using MPC global optimization; when the efficiency difference exceeds 10%, the single rule allocation is close enough to the optimal, and the rule algorithm with lower computational complexity is more advantageous.

[0168] Second, based on the analysis of the vehicle's longitudinal dynamics characteristics, the optimal operating point distribution of the dual-motor system under different accelerations was analyzed using the ADVISOR and CRUISE simulation platforms. The threshold of 0.9 is based on the fact that when the required acceleration exceeds 0.4g (approximately 3.92 m / s²), dynamic performance becomes the absolute dominant requirement, and the space for economic optimization is extremely small. In this case, a torque bridge-dominated strategy should be adopted entirely.

[0169] Third, based on cluster analysis of typical operating condition data from real vehicle testing, operational data of a 6×4 electric drive axle tractor in a western province was collected. The K-means clustering algorithm was used to cluster the three-dimensional data of [vehicle speed, throttle opening, and actual acceleration]. Results: The cluster centers were naturally distributed in four regions, corresponding to four operating condition types, with their boundary values, after statistical analysis, approximately around 0.3, 0.7, and 0.9.

[0170] Table 2 Setting Example

[0171]

[0172] calculate:

[0173]

[0174]

[0175]

[0176] Numerical meaning: On the continuous spectrum of working conditions from 0 to 1, the 0.86 biased torque bridge takes precedence;

[0177] Control strategy: The system should adopt a distribution strategy based on torque bridge.

[0178] The calculation process involves first identifying which rules are activated, then calculating the activation weight of each rule, and finally applying the rule's confidence level and... The set is then weighted and averaged to obtain the final result. value.

[0179] In this invention, These are fixed, unchanging operating condition labels in the algorithm. A set of changing weights is dynamically calculated based on real-time input. Finally calculated It is a continuous value. This fixed operating condition label + dynamic parameter activation structure is an important manifestation of the intelligence and adaptability of this invention.

[0180] 2.4 Fourth step, multi-mode adaptive torque distribution algorithm

[0181] Based on the identified operating condition type, corresponding allocation algorithms are adopted, including: a high-torque-dominant operating condition algorithm (a feasible solution under dynamic constraints), a high-speed-dominant operating condition algorithm (a feasible solution under efficiency-priority constraints), a comprehensive operating condition global optimization algorithm (a globally optimal solution under multiple objectives and constraints), and a real-time optimization algorithm based on complementary efficiency MAP (a real-time approximate optimal solution when computational resources are limited). These four algorithms are complementary, and the calculated result is the torque allocation between the torque bridge and the speed bridge, essentially representing different solutions to a multi-dimensional optimization problem. This allocation algorithm outputs the optimal allocated torque through operating condition identification → algorithm selection → parallel computation → result fusion → constraint arbitration. The calculation process is as follows: Figure 7 .

[0182] The algorithm design is as follows:

[0183] 2.4.1 Algorithm for High Torque Dominant Operating Conditions

[0184] The rule of prioritizing torque bridge is adopted:

[0185] Torque distribution by the torque bridge:

[0186] Speed ​​bridge torque distribution:

[0187] in Torque required for the entire vehicle; Distribute torque to the torque bridge; Distribute torque to the speed bridge.

[0188] For dynamic adjustment coefficients:

[0189]

[0190] In the formula, Basic allocation coefficient, This represents the maximum available torque of the torque bridge (limited by motor capacity, temperature, etc.). , To adjust the gain. For the desired acceleration (by (and v calculation) This is the actual acceleration (calculated by VCU).

[0191] : Total required torque The torque is proportionally allocated to the torque axle to ensure it operates within its efficient range. Remaining torque is supplemented by the speed axle to ensure total demand is met.

[0192] Dynamic adjustment coefficient Intelligent design. Three components:

[0193] 1) Basic coefficient The default torque bridge distribution ratio is set, and this invention determines it through calibration. The value ranges from 0.7 to 0.95, and the specific value needs to be calibrated according to the characteristics of the motor and the vehicle usage scenario for different powertrains.

[0194] For example: =0.8 indicates that 80% of the torque is allocated to the torque axle by default.

[0195] 2) Switch opening compensation item : Dynamically adjusts according to the intensity of the driver's intention.

[0196] : Reference switch opening (e.g., 40%), when > The driver presses the accelerator pedal deeply, increasing the torque axle ratio. Adjust the gain (e.g., 0.1-0.3).

[0197] Example: =0.2, =80%, =40% → Increase the allocation ratio by 0.2*(80%-40%)=8%.

[0198] 3) Acceleration deviation compensation item : Dynamically adjust based on the actual response of the vehicle

[0199] when The acceleration is insufficient and more torque is needed.

[0200] 2.4.2 Algorithm for High-Speed ​​Dominant Operating Conditions

[0201] In high-speed cruising scenarios, the core objective is to maximize fuel economy. Priority is given to ensuring that the speed bridge operates in its high-efficiency range, and as close as possible to its optimal efficiency point. Any insufficient torque is provided by the torque bridge.

[0202] Optimal allocation of speed bridge efficiency:

[0203] Optimal torque efficiency of the speed bridge:

[0204] Speed ​​bridge torque distribution:

[0205] Torque distribution by the torque bridge:

[0206] In the formula, This represents the optimal efficiency point for the speed bridge. Let the speed bridge efficiency function be... The speed bridge allocation coefficient. Distribute torque to the speed bridge, Distribute torque to the torque bridge.

[0207] The optimization objective The engineering implementation method is as follows: based on a pre-calibrated three-dimensional MAP of the speed bridge motor efficiency (each speed, torque, and motor temperature corresponds to a motor efficiency), and based on the real-time collected motor speed and temperature, the optimal efficiency is found. The maximum value corresponds to the torque.

[0208] Algorithm implementation principle: Function: Speed ​​bridge efficiency, which is the current torque T and the current speed. The function searches using a pre-calibrated three-dimensional efficiency map. `argmax`: At the current speed, scans the efficiency map, finds the operating point with the highest efficiency, and outputs the torque value at that point. .

[0209] 1) Intelligent allocation logic

[0210]

[0211]

[0212] Handling two situations:

[0213] when The required torque is less than the torque at the optimal efficiency point, proportionally. The load is allocated to the speed bridge, and the remainder to the torque bridge. The speed bridge operates in the high-efficiency range.

[0214] when The required torque exceeds the optimal efficiency point, thus limiting the speed bridge torque to... (Optimal efficiency point) The remaining torque is allocated to the torque bridge to avoid the speed bridge operating in the inefficient high torque region.

[0215] 2) Allocation coefficient Design

[0216] This invention determines through calibration The torque ranges from 0.6 to 0.9, with adjustments made according to calibration for different powertrains. During high-speed cruising, most of the torque should be provided by the speed axle, with minor adjustments based on factors such as vehicle speed and battery SOC.

[0217] 3) Application Examples

[0218] Scenario: High-speed cruising, v=100km / h =2000 Nm;

[0219] Given: =4000rpm, =1500 Nm;

[0220] set up: = 0.8;

[0221] calculate:

[0222] = 0.8 = 1600Nm;

[0223] But 1600 Nm > (1500Nm);

[0224] therefore: = min(1600, 1500) = 1500Nm;

[0225] = 2000 - 1500 = 500Nm;

[0226] Result: The speed bridge operates at its optimal efficiency point (1500 Nm), while the torque bridge provides the remaining 500 Nm.

[0227] 2.4.3 Global Optimization Algorithm for Combined Working Conditions

[0228] Under complex and varied operating conditions such as suburban roads, mountain roads, and slippery roads, simple rule allocation is insufficient to achieve global optimization; multiple objectives, including energy consumption, responsiveness, and smoothness, must be considered simultaneously. This invention employs a model predictive control (MPC) framework for multi-objective optimization to minimize the system's total energy consumption.

[0229] The minimum energy of the system is:

[0230]

[0231] The power output of the power battery when driving the torque bridge; The power output of the power battery when driving the speed axle; The minimum output power of the power battery is the core optimization target.

[0232] Discretizing the above equation yields:

[0233]

[0234] The minimum power output of the discrete power battery; Indexed by time step; To predict the time domain length, and to consider the total number of future discrete time steps in MPC; λ: energy consumption weighting coefficient, usually set to 1; The weighting coefficients for the torque tracking error term; At any moment Total torque required for the entire vehicle; At any moment The actual total output torque of the dual motors.

[0235] Energy minimization term: λ * .

[0236] Torque tracking item: This ensures that the actual output torque of the dual motors tracks the total required torque.

[0237] To prevent sacrificing power response for energy saving, the scalability options can also include smoothness penalties, temperature penalties, etc. (e.g.: ).

[0238] Optimization must be performed within the limits of the system's physical constraints and safety rules, therefore the constraints are:

[0239] 1) Torque upper and lower limits:

[0240]

[0241]

[0242] This is the minimum output torque of the torque bridge. This represents the maximum output torque of the torque bridge.

[0243] 2) Torque balance:

[0244]

[0245] 3) Battery power:

[0246]

[0247] This refers to the total output power of the power battery. This is the maximum output power of the power battery.

[0248] 4) Battery SOC

[0249]

[0250]

[0251] This refers to the state of charge of the power battery. This represents the minimum state of charge of the power battery. This represents the maximum state of charge of the power battery.

[0252] Minimum power output of the discrete power battery The solution is obtained using a Model Predictive Control (MPC) framework. The state-space model is as follows:

[0253] Equations of state:

[0254] This is a linear model used to predict the future behavior of a system, representing the state at the current moment. Add control input Together, they determine the state of the next moment. .

[0255] in, This represents the state vector (the system's "memory") at time k. This represents the state vector at time k+1; The control input applied at time k; A is the state transition matrix; B is the control input matrix.

[0256] State vector = ;

[0257] SOC: Battery state of charge, reflecting the state of the energy source;

[0258] The torque bridge speed determines the motor's operating point;

[0259] The speed of the speed bridge determines the operating point of the motor;

[0260] : The torque bridge command from the previous moment, used for continuity;

[0261] : The speed bridge instruction from the previous moment, used for continuity.

[0262] Output equation:

[0263] Let C represent the output vector at time k; C is the output matrix.

[0264] 5) Make Minimum control input u = The control input u is the torque command, which is the direct object of the optimization solution, namely the increment of the torque of the two motors relative to the previous moment.

[0265] This is the increment of the torque command for the torque bridge; Increment of speed bridge torque command

[0266] The rolling optimization problem of model predictive control (MPC) (minimizing over N future time steps) This can be transformed into standard QP form and solved online using a quadratic programming (QP) solver. The matrix representation is as follows:

[0267]

[0268] H is the Hessian matrix; F is the gradient vector.

[0269] Output result:

[0270] Optimal control sequence (n future steps): u = Output the current torque distribution: .

[0271] 2.4.4 Real-time Optimization Algorithm Based on Complementary Efficiency Map (MAP)

[0272] For the vehicle control unit (VCU) with limited computing resources, real-time optimization based on complementary efficiency map (MAP) is adopted. The complementarity is reflected in the fact that when the torque bridge efficiency is high, more resources are allocated to the torque bridge; when the speed bridge efficiency is high, more resources are allocated to the speed bridge.

[0273] Define the overall system efficiency:

[0274]

[0275]

[0276] In the formula, k represents the torque distribution coefficient (for example, when k=0.3, the torque bridge bears 30% of the total required torque). Represents the overall efficiency of the system at a certain moment; This represents the output torque of the torque bridge at a certain moment; This represents the output torque of the speed bridge at a certain moment; This represents the output speed of the torque bridge at a certain moment; This represents the speed bridge output rotational speed at a certain moment; This represents the torque bridge efficiency at a certain moment. This represents the bridge efficiency at a certain moment. (Formula)

[0277] In the formula The efficiency is obtained by looking up a 3D efficiency MAP table for the motor (three dimensions: torque, speed, and temperature, each corresponding to an efficiency).

[0278] Numerator: Total mechanical power output of the two bridges; Denominator: Electrical power output of the power battery (converted using efficiency η); Ratio Overall system efficiency.

[0279] Efficiency acquisition methods:

[0280] , , )

[0281] , , )

[0282] Lookup table structure:

[0283] Dimension 1: Torque (e.g., 0, 100, 200, ...) Nm)

[0284] Dimension 2: Rotational speed (e.g., 0, 1000, 2000, ..., max_rpm)

[0285] Dimension 3: Temperature (e.g., 0, 40, 80, 120 °C)

[0286] Numerical value: Efficiency value (0%-100%)

[0287] The optimal allocation coefficient k is searched using the mature univariate function extremum search algorithm, the "golden section method," to make... maximum:

[0288]

[0289] The algorithm outputs a quadruple result:

[0290] Real-time optimization results =

[0291]

[0292]

[0293]

[0294] 2.5 Fifth step, adaptive learning

[0295] Establish an online learning module to update control parameters in real time:

[0296]

[0297] In the formula, The parameters to be learned at step k are the variables to be adjusted (e.g., the total mass of the vehicle, m). veh (Load changes, efficiency MAP offset, slope changes, etc.); γ is the learning rate, which controls the magnitude of each update step to avoid overshoot or slow convergence. This refers to the actual performance indicators (such as energy consumption) measured at step k. The model prediction performance at step k. For the performance J at step k, the parameters are... The gradient.

[0298] Output content: The updated parameter vector is used for VCU computation at the next time step. The algorithm runs continuously at each time step, forming a parameter sequence. Through adaptive learning, the VCU updates in real time at each time step based on the current vehicle state parameters (vehicle gross weight, speed, throttle opening, etc.) using an embedded learning algorithm. All operating conditions encountered by the vehicle during normal driving become learning data for the algorithm. The algorithm learns in a real, continuous flow of operating conditions, and the system always tracks the vehicle's optimal state, rather than a previously calibrated optimal state that may be outdated.

[0299] The core innovation of this invention lies in abandoning the traditional average or fixed-ratio torque distribution strategy and proposing a dynamic, real-time torque distribution method based on the collaborative optimization of motor efficiency characteristics, aiming to minimize the overall energy consumption of the system. Simultaneously, it proposes a vehicle-axle architecture design concept with complementary efficiency maps to extend the system's efficient operating range from a physical perspective.

[0300] 1. Hardware architecture design for complementary high-efficiency zones:

[0301] For the first time, a differentiated design specification for torque bridges and speed bridges was systematically proposed. Starting from the energy efficiency optimization target of electric drive bridge system, the high-efficiency operating range of electric drive bridges was broadened from a physical level.

[0302] 2. Core Concept Innovation: From "Distributing Torque" to "Distributing Efficiency"

[0303] The traditional approach is to simply distribute the torque in a fixed ratio or on an average basis, with the goal of meeting the total torque demand.

[0304] The core idea of ​​this patent is to transform the problem of allocating total required torque into a collaborative optimization problem of how to ensure that the two motors on the front and rear axles operate within their respective ranges of highest overall efficiency (or lowest energy consumption). It seeks to achieve the optimal global system efficiency, rather than the optimal efficiency of a single motor or a localized area.

[0305] 3. Innovative Methodological Architecture: Dynamic Allocation Based on "Global Optimization + Local Matching"

[0306] The patented method likely contains a two-layer or composite decision structure:

[0307] Upper layer (global efficiency optimizer): Based on the vehicle's current state (vehicle speed, total torque demand), battery state, and the high-efficiency zone MAPs of the front and rear motors, it calculates a theoretically optimal torque distribution ratio using an optimization algorithm (possibly similar to the model predictive control or instantaneous optimization algorithm you mentioned earlier). The objective function of this optimization is to minimize the total energy consumption of the dual-motor system.

[0308] Lower layer (complementary high-efficiency region matching): The key lies in the word "complementary." The algorithm will intelligently determine:

[0309] When the total torque demand is low, it may be entirely carried by a single motor in the high-efficiency zone (such as the rear axle main drive motor), while the other motor is on standby, thus avoiding both motors operating in the low-efficiency zone.

[0310] When the total demand torque increases and enters the low-to-medium efficiency range of a single motor, another motor is started, and the torque distribution is adjusted so that both motors can operate in their respective medium-to-high efficiency ranges as much as possible, thereby improving overall efficiency through "complementarity".

[0311] Under conditions of high torque demand, such as rapid acceleration and hill climbing, the torque distribution is optimized to balance efficiency and heat generation while meeting the constraints of power performance.

[0312] 4. Technological Innovations (Specific Examples)

[0313] Real-time lookup and calculation based on efficiency MAP: The system learns the efficiency characteristic curves (MAP diagrams) of the motors before and after the process by either built-in or online methods, and uses these as the fundamental basis for allocation decisions.

[0314] Multi-constraint dynamic optimization: The allocation process not only considers efficiency but also integrates:

[0315] Vehicle dynamics constraints (such as anti-skid and axle load transfer);

[0316] Electrical heating safety constraints (motor temperature, controller temperature);

[0317] Battery power limitation;

[0318] Driving comfort constraints (avoiding abrupt assignment changes);

[0319] Adaptive and learning capabilities: The system has the ability to learn online about changes in motor efficiency MAP and vehicle parameters (such as mass and resistance), so that the allocation strategy remains optimal throughout the entire process.

[0320] 5. System-level innovation and its advantages

[0321] By consistently pursuing overall system efficiency, we directly reduce vehicle energy consumption and increase driving range.

[0322] Extended component life: Optimized allocation can prevent a single motor from being continuously overloaded or operating in a high-temperature and inefficient area, balancing the thermal load and mechanical wear of the dual-motor electric drive system.

[0323] Enhanced driving adaptability: The strategy can adapt to different operating conditions (city, highway, hill climbing) and different loads (no load, full load) to achieve energy efficiency optimization under all operating conditions.

[0324] Highly practical for engineering applications: This method is typically implemented as a software algorithm, which can tap the energy-saving potential of existing dual-motor systems without increasing the computing power and cost of the VCU.

[0325] In summary, the most fundamental innovation of this patent lies in proposing and implementing a dynamic torque distribution intelligent decision-making mechanism based on the complementary efficiency characteristics of dual motors, with the goal of achieving "real-time optimal global system efficiency." It elevates drive torque distribution from a relatively static, rule-based control problem to a dynamic, optimization-model-based system-level energy efficiency management problem, thus realizing the optimal control method.

[0326] The beneficial effects of this invention are:

[0327] The beneficial effects of this invention are significant and comprehensive, mainly reflected in three aspects: breakthroughs in technical performance, improved economic benefits, and enhanced system reliability. Compared with existing dual-electric drive bridge systems with similar efficiency characteristics and simple 50%:50% allocation strategies, this invention achieves a qualitative leap through the synergistic innovation of hardware architecture and intelligent control algorithms.

[0328] 1. Breakthrough in core technology performance

[0329] 1.1 Significantly reduced energy consumption across all operating conditions (core benefit)

[0330] Through the complementary high-efficiency zone design of torque bridge and speed bridge and the intelligent allocation algorithm based on multi-parameter fusion, the system can always allocate the main load to the drive bridge with the highest efficiency under the current working conditions.

[0331] Data support: Under user operating conditions or comprehensive operating cycle such as CHTC, the energy consumption of the whole vehicle can be reduced.

[0332] When climbing at low speeds, the high-torque-efficiency torque bridge dominates; when cruising at high speeds, the high-speed-efficiency speed bridge dominates; under combined operating conditions, the algorithm solves the global optimal allocation point in real time to avoid both motors operating in the inefficient zone.

[0333] 1.2 Simultaneous optimization of power performance and drivability

[0334] This invention overcomes the traditional contradiction between economy and power.

[0335] Enhanced power response: During rapid acceleration (high β, high da / dt), it is identified as a "transient acceleration" or "high torque-dominated" condition, and the system prioritizes ensuring the rapid torque response of the torque bridge.

[0336] Enhanced high-speed power reserves: Due to the independent optimization of high-speed performance by the speed axle, the vehicle's maximum speed can be increased, and its high-speed acceleration capability is stronger.

[0337] Significantly enhanced climbing ability: The high reduction ratio of the torque axle and the constant torque range design of the motor increase the vehicle's maximum climbing gradient, meeting the needs of complex road conditions. (This can be deleted.)

[0338] 1.3 Horizontal Leap in Intelligent and Adaptive Control

[0339] Precise operational condition perception: A three-in-one recognition mechanism integrating vehicle speed (v), throttle opening (ß), and acceleration (a) improves the accuracy of judging driver intent and actual load compared to strategies that rely solely on vehicle speed.

[0340] Forward-looking optimization capability: By adopting the model predictive control (MPC) framework, it optimizes energy consumption based on the prediction time domain, realizing the leap from "instantaneous optimal" to "short-term global optimal", balancing smoothness and economy.

[0341] Self-learning and evolution: The adaptive learning module can correct the efficiency MAP and allocation parameters online to adapt to vehicle aging, load changes and driver habits, so that the system can maintain peak performance throughout its entire life cycle.

[0342] 2. Economic benefits and industrial value

[0343] Direct energy-saving benefits: Based on 100,000 kilometers of annual operation for commercial vehicles and an electricity price of RMB 1 / kWh, a 10% reduction in energy consumption can save approximately RMB 15,000 to 25,000 in electricity costs annually.

[0344] Extended component lifespan: Reasonable load distribution avoids continuous inefficient operation and overheating of the motor, which is expected to extend the lifespan of the electric drive system by more than 20% and reduce maintenance costs.

[0345] 3. Enhanced system reliability and security

[0346] 3.1 Improved thermal management reliability

[0347] The intelligent load allocation strategy is essentially a "heat load management strategy." By actively directing the main load to the bridge with the highest current efficiency through algorithms, the heat loss of the dual bridges is reduced, the risk of motor demagnetization at high temperatures is decreased, and the system's durability under extreme operating conditions is improved.

[0348] 3.2 Strong fault tolerance and degradation operation capabilities

[0349] Single-axle fail-safe mode: When either the torque axle or the speed axle fails, the system seamlessly switches to single-axle drive mode and intelligently adjusts the vehicle's power limit based on the capacity of the remaining healthy axle (torque axle or speed axle) to ensure the vehicle's basic mobility and improve safety.

[0350] Multi-level constraint protection: The algorithm incorporates multiple hard and soft constraints such as motor torque / speed / power, battery power, and temperature to ensure that any allocation command is within the physical safety boundary, preventing system overload from the source.

[0351] 3.3 Improve smoothness and ride comfort

[0352] MPC's rolling optimization strictly limits the rate of change of torque distribution commands, avoiding vehicle pitch shock caused by drastic switching of torque distribution between the front and rear axles, thus improving the ride quality.

Claims

1. A dual-motor system for commercial vehicles based on complementary high-efficiency zones, characterized in that: The dual-motor system includes a torque bridge and a speed bridge. The torque bridge is equipped with a high torque density motor and a large reduction ratio reducer, with its high efficiency range concentrated in the low-speed, high-torque region. The low speed and high torque refer to 0%-60% of the rated speed and 30%-100% of the peak torque. The speed bridge is equipped with a high power density motor and a small reduction ratio reducer, with its high efficiency range concentrated in the high-speed, low-to-medium torque region. The high speed and low-to-medium torque region refer to 30%-100% of the rated speed and 10%-60% of the peak torque. The high efficiency ranges of the two bridges complement each other on the speed-torque plane, jointly covering the entire operating range of the vehicle. When high torque is required, the torque bridge shares a larger share of the output power; when high vehicle speed is required, the speed bridge shares a larger share of the output power, always maintaining a higher output power in the high efficiency range.

2. A torque distribution control method for a dual-motor system in a commercial vehicle based on complementary high-efficiency zones, characterized by: Includes the following steps: The first step is multi-source signal acquisition and fusion processing. The signals transmitted to the vehicle control unit VCU include: vehicle speed v, accelerator opening degree , vehicle longitudinal acceleration ɑ, motor speed , temperature T, battery SOC% state parameters; As a state vector, the original signal is denoised and fused by calling Kalman filtering algorithm from the application layer software of the vehicle control unit VCU, further denoising and improving the signal quality, thereby improving the robustness and precision of the control system, and the five equations of the Kalman filtering algorithm are as follows: State prediction equation: Error covariance prediction equation: Kalman gain prediction equation: State update equation: Error covariance update equation: wherein, is a state vector, is an observation vector, F, H are state transition and observation matrices, P is an estimation error covariance matrix, Q, R are process noise and measurement noise covariance matrices, is a Kalman gain; k is the time step, k|k means that the information up to and including the latest observation data at time k is used at time k, k|k-1 means that the information at time k-1 is given at time k-1, and k-1|k-1 means that the information at time k-1 has been fused with the latest observation data at time k-1. The second step is to calculate the total torque requirement of the entire vehicle. Total torque required for the whole vehicle: Driver's required torque In the formula, The basic torque table is a static mapping based on a two-dimensional lookup table, defining the basic torque. With switch opening Speed The relationship, that is, each ( The coordinates correspond to a basic torque. The values ​​in this table are determined through vehicle calibration and are placed in the program of the vehicle control unit (VCU). During calculation, the values ​​are looked up and called in real time. 1) Dynamic compensation term , electric door opening change compensation, capture the transient intention of the driver: : the difference between the original electric door opening and the filtered opening: When the driver quickly presses the pedal: The compensation term is positive, increasing the torque output and improving the response speed. When the driver quickly releases the pedal: The compensation term is negative, quickly reducing the torque, enhancing energy recovery; 2) Parameters : proportional gain, value 0.1-0.5, can be adjusted according to specific working conditions, and response and smoothness should be considered during calibration; 3) a differential compensation term represents a potentiometer opening rate of change compensation, predicts the intention trend of the driver, is the potentiometer opening rate of change: When the driver steps on the pedal sharply: Very large and positive, the compensation term adds torque, providing acceleration. When the driver slowly presses the pedal: Smaller, the compensation term is smaller, and smooth acceleration is ensured; Parameters : differential gain, 0.05-0.2, can be adjusted according to specific working conditions; Inertia compensation torque In the formula, is the equivalent rotational inertia, is the tire radius, is the total speed ratio, is the motor inertia, is the total mass of the vehicle; Driving resistance torque wherein for air resistance, for air density, for wind resistance coefficient, A is the effective windward area of the cab, for vehicle speed; for the rolling resistance, for the total vehicle weight, for the acceleration of gravity, for the rolling resistance coefficient, and θ for the road slope; For slope resistance; The third step is the intelligent working condition recognition algorithm. Intelligent operating condition recognition algorithm: Based on vehicle speed v, throttle opening ß, and vehicle acceleration α, a method combining fuzzy logic and rule-based judgment is used to identify the operating conditions. The corresponding torque logic is as follows: 1) Define the fuzzy set of input variables: a) Vehicle speed v: {very low, low, medium, high, very high}; b) Switch opening ß: {very small, small, medium, large, very large}; c) Acceleration Demand Index : {very small, small, medium, large, very large}; 2) Based on vehicle dynamics theory, motor efficiency MAP characteristics, whole vehicle simulation analysis results, and real vehicle calibration test data, a fuzzy rule base is comprehensively formulated, which includes several rules corresponding to several working conditions; Meanwhile, based on the statistical analysis of historical data and the energy consumption calculation results based on simulation optimization, the confidence level of each rule is determined. The confidence level represents the determinism of the rule in actual vehicle operation. When multiple rules conflict, the rule with higher confidence level has higher weight. 3) Finally, the weighted average method is used for defuzzification, and the working condition type and confidence level are output: wherein, is the rule activation weight, representing the minimum membership degree among the membership values in the current input variable, at which the rule is triggered; Mode i is the numerical label corresponding to each fuzzy rule, similar to an "identity card number", which maps different working conditions to a specific position in the continuous interval [0, 1]. The value is pre-set and reflects the degree of inclination of torque distribution. The setting principle of the value is: Orderliness: The operating conditions are arranged in the order from the speed bridge to the torque bridge; continuity, of adjacent operating conditions values close to, smooth transition; Non-overlapping, different operating conditions Values have reasonable spacing; In terms of physical meaning, the magnitude of the value reflects the degree of torque distribution tendency. The setting also needs to combine with simulation analysis and real vehicle calibration adjustment, so that the error of the set value and the calculated value meets the control requirements, which is a repeated iteration process, and finally in the vehicle control unit VCU, The value is stored in the form of a lookup table or a constant. Step 4: Multi-mode adaptive torque distribution algorithm Based on the identified operating condition type, corresponding allocation algorithms are adopted, including: high torque-dominant operating condition algorithm, high speed-dominant operating condition algorithm, comprehensive operating condition global optimization algorithm, and real-time optimization algorithm based on complementary efficiency MAP. The four algorithms are complementary. The calculation result is the torque allocation between the torque bridge and the speed bridge. The allocation algorithm outputs the optimal allocated torque through operating condition identification → algorithm selection → parallel calculation → result fusion → constraint arbitration.

3. The method according to claim 2, characterized in that: The algorithm for high-torque-dominant operating conditions adopts a torque bridge priority rule: Torque bridge distributes torque: Speed ​​bridge torque distribution: in Torque required for the entire vehicle; Distribute torque to the torque bridge; Distribute torque to the speed bridge; For dynamic adjustment coefficients: wherein is the base allocation factor, is the current maximum torque available for the torque bridge, limited by motor capability, temperature, , is the adjustment gain, is the desired acceleration, calculated from and v, is the actual acceleration; : The total demand torque is proportionally distributed to the torque bridge, ensuring that the torque bridge is in its high efficiency region. : The total demand torque is proportionally distributed to the torque bridge, ensuring that the torque bridge is in its high efficiency region. Remaining Torque: Supplemented by the speed bridge to ensure total demand is met.

4. The torque distribution control method for a commercial vehicle dual-motor system based on complementary high-efficiency zones according to claim 3, characterized in that: The dynamic adjustment coefficient The design includes: 1) Base coefficient , set default torque bridge distribution ratio, determined by calibration 0.7-0.95, different powertrain specific according to motor characteristics and vehicle use scene calibration; 2) a potentiometer opening compensation term : dynamic adjustment according to the driver's intention strength; : Reference switch opening, when > The driver presses the accelerator pedal deeply, increasing the torque axle ratio. Adjust the gain; 3) Acceleration deviation compensation item : Dynamically adjust based on the actual response of the vehicle when The acceleration is insufficient and more torque is needed.

5. The torque distribution control method for a commercial vehicle dual-motor system based on complementary high-efficiency zones according to claim 2, characterized in that: The algorithm for high-speed dominant operating conditions employs optimal allocation of speed bridge efficiency. Optimal torque efficiency of the speed bridge: Speed ​​bridge torque distribution: Torque distribution by the torque bridge: In the formula, This represents the optimal efficiency point for the speed bridge. Let the speed bridge efficiency function be... Assigning coefficients to the speed bridge. Distribute torque to the speed bridge, Distribute torque to the torque bridge; The optimization objective The engineering implementation method is as follows: based on a pre-calibrated three-dimensional map of the speed bridge motor efficiency, and according to the real-time collected motor speed and temperature, find the optimal efficiency. The maximum value corresponds to the torque.

6. The torque distribution control method for a commercial vehicle dual-motor system based on complementary high-efficiency zones according to claim 5, characterized in that: the algorithm... Implementation principle: Function: Speed ​​bridge efficiency, which is the current torque T and the current speed. The function searches a pre-calibrated three-dimensional efficiency map. `argmax`: at the current speed, it scans the efficiency map to find the operating point with the highest efficiency and outputs the torque value at that point. This torque value is... ,include: 1) Intelligent allocation logic Handling two situations: when The required torque is less than the torque at the optimal efficiency point, proportionally. The load is allocated to the speed bridge, and the remainder to the torque bridge. The speed bridge operates in the high-efficiency range. when The required torque exceeds the optimal efficiency point, limiting the speed bridge torque to... The remaining torque is allocated to the torque axle, preventing the speed axle from operating in the inefficient high-torque range; 2) Allocation coefficient Design Calibration determination The torque is 0.6-0.9, and different powertrains are adjusted according to the calibration. During high-speed cruising, most of the torque should be provided by the speed axle, and it is finely adjusted according to vehicle speed and battery SOC factors.

7. The torque distribution control method for a commercial vehicle dual-motor system based on complementary high-efficiency zones according to claim 2, characterized in that: The comprehensive operating condition global optimization algorithm uses a model predictive control (MPC) framework for multi-objective optimization to minimize the total system energy consumption. The minimum energy of the system is: The power output of the power battery when driving the torque bridge; The power output of the power battery when driving the speed axle; The minimum output power of the power battery is the core optimization target; Discretizing the above equation yields: The minimum power output of the discrete power battery; Indexed by time step; To predict the length of the time domain, MPC considers the total number of future discrete time steps; λ: Energy consumption weighting coefficient, usually set to 1; The weighting coefficients for the torque tracking error term; At any moment Total torque required for the entire vehicle; At any moment The actual total output torque of the dual motors; Energy minimization term: λ * ; Torque tracking item: This ensures that the actual output torque of the dual motors tracks the total required torque. Optimization must be performed within the limits of the system's physical constraints and safety rules, therefore the constraints are: 1) Torque upper and lower limits: This is the minimum output torque of the torque bridge. This represents the maximum output torque of the torque bridge. 2) Torque balance: 3) Battery power: This refers to the total output power of the power battery. This represents the maximum output power of the power battery. 4) Battery SOC This refers to the state of charge of the power battery. This represents the minimum state of charge of the power battery. This represents the maximum state of charge of the power battery. Minimum power output of the discrete power battery The solution is obtained using a Model Predictive Control (MPC) framework. The state-space model is as follows: Equations of state: It is a linear model used to predict the future behavior of a system, representing the state at the current moment. Add control input Together they determine the state of the next moment. ; in, This represents the state vector at time k; This represents the state vector at time k+1; The control input applied at time k; A is the state transition matrix; B is the control input matrix; State vector = ; SOC: Battery state of charge, reflecting the state of the energy source; The speed of the torque bridge determines the operating point of the motor; The speed of the speed bridge determines the operating point of the motor; : The torque bridge command from the previous moment, used for continuity; : The speed bridge command from the previous moment, used for continuity; Output equation: Let represent the output vector at time k; C is the output matrix. 5) Make Minimum control input u = The control input u is the torque command, which is the direct object of optimization solution, that is, the increment of the torque of the two motors relative to the previous moment. This is the increment of the torque command for the torque bridge; Increment of speed bridge torque command The rolling optimization problem of model predictive control (MPC) is to minimize the time step over the next N time steps. It can be transformed into standard QP form and solved online using a quadratic programming QP solver. The matrix representation is as follows: H is the Hessian matrix; F is the gradient vector; Output result: The optimal control sequence is the sequence of the next N steps: u = Output the current torque distribution: . Torque tracking item: This ensures that the actual output torque of the dual motors tracks the total required torque.

8. The torque distribution control method for a commercial vehicle dual-motor system based on complementary high-efficiency zones according to claim 7, characterized in that: To prevent sacrificing power response for energy saving, torque tracking is implemented. Scalable, can incorporate smoothness penalties and temperature penalties: ).

9. The torque distribution control method for a commercial vehicle dual-motor system based on complementary high-efficiency zones according to claim 2, characterized in that: The complementarity of the real-time optimization algorithm based on complementary efficiency MAP is reflected in the fact that when the torque bridge efficiency is high, more torque bridge is allocated. When the speed bridge is efficient, more resources should be allocated to the speed bridge; Define the overall system efficiency: In the formula, k represents the torque distribution coefficient; Represents the overall efficiency of the system at a certain moment; This represents the output torque of the torque bridge at a certain moment; This represents the output torque of the speed bridge at a certain moment; This represents the output speed of the torque bridge at a certain moment; This represents the speed bridge output rotational speed at a certain moment; This represents the torque bridge efficiency at a certain moment. Represents the speed-bridge efficiency at a certain moment; In the formula The efficiency is obtained by looking up a 3D efficiency MAP table for the motor, which has three dimensions: torque, speed, and temperature, each corresponding to an efficiency. Numerator: Total mechanical power output of the two bridges; Denominator: Electrical power output of the power battery. The optimal allocation coefficient k is searched using the mature univariate function extremum search algorithm, the "golden section method," to make... maximum: The algorithm outputs a quadruple result: Real-time optimization results = 。 10. The torque distribution control method for a commercial vehicle dual-motor system based on complementary high-efficiency zones according to claim 2, characterized in that: The method also includes: a fifth step, adaptive learning. Establish an online learning module to update control parameters in real time: In the formula, The parameters to be learned at step k, i.e., the variables to be adjusted, include: the total mass of the vehicle, m. veh That is, load changes, efficiency MAP offset, and slope changes; γ is the learning rate, which controls the magnitude of each update step to avoid overshoot or slow convergence. The actual performance indicators measured at step k include energy consumption. The model prediction performance at step k is... For the performance J at step k, the parameters are... The gradient; Output content: The updated parameter vector is used for VCU computation at the next time step. The algorithm runs continuously at each time step, forming a parameter sequence. Through adaptive learning, the VCU updates in real time at each time step based on the current vehicle state parameters, including the vehicle's total mass, speed, and throttle opening, using an embedded learning algorithm. All operating conditions encountered by the vehicle during normal driving are the learning data for the algorithm. The algorithm learns in a real and continuous flow of operating conditions, and the system always tracks the vehicle's optimal state, rather than the previously calibrated optimal state that may be outdated.

Images