An adaptive weight dynamic programming hybrid vehicle train energy management method

Abstract

The application relates to the technical field of hybrid vehicle train energy management, and particularly discloses a hybrid vehicle train energy management method based on adaptive weight dynamic programming, which comprises the following steps: obtaining slope information of a vehicle by using a terminal and a map of a vehicle networking service system; calculating the maximum acceleration and the minimum acceleration of the vehicle on each road section, and iteratively obtaining the feasible region of all road sections in front; determining the vehicle speed as a state variable of a dynamic programming algorithm, and establishing a pre-planning cost function model; performing one-dimensional dynamic programming on the vehicle speed; updating the adaptive weight and the cost function; performing dynamic programming on the vehicle speed and the SOC (State of Charge); executing the planned vehicle speed and the SOC, and performing optimal energy efficiency cruise control; the application considers adaptive slope working conditions, takes into account multi-target performance, cooperatively plans to improve overall performance, and improves the consistency of driving smoothness and power response.

Description

Technical Field

[0001] This invention belongs to the field of hybrid electric vehicle train energy management technology, specifically relating to an adaptive weighted dynamic programming method for hybrid electric vehicle train energy management. Background Technology

[0002] With increasingly stringent regulations on energy conservation and emission reduction for automobiles and trains, and the growing sensitivity of logistics transportation to operating costs, the fuel economy requirements for traditional fuel-powered automobile trains are becoming increasingly stringent. Hybrid electric vehicle (HEV) trains, with their superior fuel economy and low emissions, have become the mainstream choice for the low-carbon transformation of automobile trains. HEV trains are widely used in long-distance logistics, engineering transportation, and urban delivery, operating under conditions characterized by heavy loads, long operating hours, and complex road conditions. Data shows that fuel consumption of HEV trains on mountainous roads is significantly higher than on flat roads. Gradient changes not only directly affect vehicle power demand but also determine energy recovery efficiency; therefore, the effective utilization of gradient information has become a key breakthrough in optimizing energy management strategies.

[0003] Currently, research on hybrid vehicle energy management strategies that integrate road information ahead of the vehicle primarily utilizes algorithms such as dynamic programming and model predictive control to optimize the energy allocation of the vehicle's current and future power components. However, these methods have limitations: the weight coefficients of each dimension of the cost function in dynamic programming algorithms are generally set to fixed values, and the performance of fixed weight coefficients varies under different terrain conditions, resulting in poor adaptability; because the objective function weights are fixed, it is impossible to dynamically balance multiple objectives based on the slope. Therefore, to address the technical shortcomings of existing hybrid vehicle train energy management strategies, it is necessary to design an adaptive weighted dynamic programming method for hybrid vehicle train energy management to solve the problem of fixed weights in traditional dynamic programming. Summary of the Invention

[0004] The purpose of this invention is to provide an adaptive weighted dynamic programming method for hybrid vehicle train energy management. By collecting information on the road ahead, vehicle status, and driving needs, and combining the terrain slope to dynamically adjust the objective function weights and cost function constraints, a two-dimensional dynamic programming algorithm is used to plan the future vehicle speed and battery SOC trajectory. This achieves coordinated optimization of vehicle speed planning and energy allocation in the future time domain, balances performance requirements under different operating conditions, and greatly improves energy utilization efficiency.

[0005] To address the aforementioned problems in the existing technology, the technical solution adopted in this invention is: an adaptive weighted dynamic programming method for energy management of hybrid electric vehicles, comprising the following steps:

[0006] S1. Road network reconstruction: Obtain the slope information of the road ahead using the terminal and map of the vehicle network service system;

[0007] S2. Feasible Domain Calculation: Based on the road gradient ahead, the driver's set cruise speed, and the acceptable speed fluctuation threshold, the planned speed feasible domain and SOC feasible domain are calculated in combination with vehicle dynamics formulas and battery charging and discharging formulas.

[0008] S3. Establishment of the pre-planning cost function model: Determine the vehicle speed as the state variable of the dynamic programming algorithm and establish the pre-planning cost function model;

[0009] S4. One-dimensional dynamic programming of vehicle speed: Planning within the vehicle speed range, the output provides adaptive optimization information for subsequent two-dimensional planning. The DP algorithm optimization utilizes the state transition equation to iteratively calculate within the feasible speed region in step S2. The planned speed trajectory is denoted as... V pre ;

[0010] S5. Adaptive Weight and Cost Function Update: Based on the pre-planned speed trajectory obtained in step S4, generate speed change sequences and cruise deviation sequences, and calculate reference coefficients accordingly. These coefficients are used to proportionally update the speed change weights and cruise deviation weights in the pre-planned penalty function. R2, R3 At the same time, the target for deviation will be changed from the cruise control set speed to... V pre This results in an updated penalty function;

[0011] S6. Dynamic programming of vehicle speed and SOC: Using vehicle speed and SOC as state variables, under the constraints of the feasible region of vehicle speed and the feasible region of SOC constructed in step S2, two-dimensional dynamic programming is used to find the optimal result by using the updated penalty function to obtain the final vehicle speed sequence and the optimal SOC sequence for subsequent execution control.

[0012] S7. Planning vehicle speed and SOC execution: The optimal vehicle speed sequence and SOC sequence determined in step S6 are output to the execution end for predictive energy management of the hybrid electric vehicle train; the motor output torque is determined by the planned SOC sequence, the actual required torque is subtracted from the motor output torque to obtain the engine torque, and the engine torque is controlled by a PID controller to achieve the planned target speed tracking and perform optimal energy efficiency cruise control.

[0013] Preferably, the slope information in step S1 is the slope 6 meters in front of the vehicle. km Road gradient information, based on every 300 meters... m The average slope sequence within the range is calculated and used as input for subsequent steps.

[0014] Preferably, the vehicle dynamics formula in step S2 is:

[0015] ;

[0016] in,T tq To output torque to the vehicle's powertrain. η T For the mechanical efficiency of the transmission system. i g This refers to the gear ratio of the transmission. i 0 Main reducer transmission ratio, f The rolling resistance coefficient, C D The air drag coefficient, A For windward area, α For road slope, u α For driving speed, m For the total load capacity of the vehicle, a To accelerate the vehicle, G To accelerate the vehicle from a standstill to 100 km / h The ratio of the average acceleration to the gravitational acceleration during the process.

[0017] Preferably, the construction rule for the planned vehicle speed feasible domain in step S2 is as follows: based on the maximum engine torque, motor positive torque and maximum motor negative torque of the vehicle power system, calculate the maximum acceleration and minimum acceleration reached by the vehicle in each road segment, start the iterative calculation from the cruising speed at the planning starting point to obtain the vehicle speed achievable domain, and iteratively obtain the vehicle speed feasible domain for all road segments ahead.

[0018] The construction rules for the SOC feasible region in step S2 are as follows: During forward calculation, the reduction in battery SOC is calculated based on the energy consumed when the motor outputs the maximum positive torque in each road segment, and the increase in battery SOC is calculated based on the energy replenished when the motor outputs the maximum negative torque in each road segment, thereby constructing the forward feasible region of SOC. The reverse derivation of the SOC feasible region is as follows: with the constraint that the endpoint SOC remains consistent with the initial SOC, the maximum positive / negative motor torque is used as input, and the SOC feasible interval is deduced from the endpoint to the starting point in reverse. During the process, the engine does not participate in torque output, resulting in the reverse feasible region of SOC. Finally, the intersection of the forward feasible region of SOC and the reverse feasible region of SOC is taken to obtain the final SOC feasible region.

[0019] Preferably, the battery charge / discharge formula for the battery SOC is:

[0020] ;

[0021] in, energy down This represents the total power of the motor under discharge conditions. energy up This refers to the total power of the motor under charging conditions. eff m For motor efficiency, effbat For battery efficiency, U cell Rated voltage, Ebat This represents the total energy of the battery.

[0022] Preferably, in step S3, the vehicle speed is set as the state variable of the dynamic programming algorithm. Simultaneously, the pre-planning cost function includes four cost items: fuel consumption, driving comfort, cruise speed deviation, and driving timeliness. Each cost item is assigned a corresponding initial weight. r 1 ~r 4 The formula for the pre-planning cost function model is as follows:

[0023] ;

[0024] in, f k+1 For the first k+1 The overall cost value, Q k+1 For the first k+1 The fuel consumption cost per unit, V k For the first k The vehicle speed, V k+1 For the first k+1 The vehicle speed, For the first k+1 The and the first k The average speed of each vehicle V CC Set the speed for cruise control. t k+1 For the first k+1 The running time of each vehicle r 1 For pre-planned fuel quantity weighting, r 2 Weighting of pre-planned vehicle speed changes, r 3 To set speed deviation weights for pre-planning and cruise control, r 4 This is a weight for the timeliness of pre-planning.

[0025] Preferably, the same state transition equation is used in both steps S4 and S6, specifically:

[0026] ;

[0027] The variable definitions defined in step S4 include: For the first k+1 The cost value per vehicle speed From uk arrive u k+1 The cost value matrix, For the first k The cost value per vehicle speed ;

[0028] The variable definitions defined in step S6 include: For the first k+1 Two-dimensional state cost values, namely vehicle speed and battery. SOC The cost values ​​of these two state variables, To be from two-dimensional state u k To two-dimensional state u k+1 The cost value matrix, For the first k The cost value of a two-dimensional state. .

[0029] Preferably, the cost function model formula in step S5 is:

[0030] ;

[0031] in, f k+1 For the first k+1 The cost value per vehicle speed Q k+1 For the first k+1 Fuel consumption at a given vehicle speed V k For the first k The vehicle speed, V k+1 For the first k+1 The vehicle speed, For the first k+1 The and the first k The average speed of each vehicle The final result obtained in step S4 k+1 A planned speed trajectory, t k+1 For the first k+1 The running time at each vehicle speed R 1 As a weighted average of fuel quantity, R 2 As the weight of the change in vehicle speed, R 3 To set the speed deviation weight with cruise control, R 4 Timeliness is a weighting factor.

[0032] Preferably, the specific steps for adaptively updating the weights of vehicle speed change and cruise set speed deviation based on the road gradient information in step S5 are as follows:

[0033] (1) For the cost of speed variation, the speed planning results will be used. V pre Differential calculations are performed to obtain the speed change sequence for each road segment. The absolute value of this sequence is then taken and normalized to obtain the adaptive reference coefficient for each road segment. P1 list ;

[0034] (2) For the cost of deviation from cruising speed, the speed planning results will be used. V pre Subtract cruise control speed V CC The speed deviation sequence of each road segment is obtained, and the absolute value and normalization operations of this sequence are performed to obtain the adaptive reference coefficient of each road segment. P2 list .

[0035] Preferably, the adaptive weight function in step S5 is constructed as follows:

[0036] ;

[0037] in, R 1 , R 4 Initial weighting coefficients r 1 , r 4 constant, R 2 , R 3 Optimize coefficients using adaptive weights P1 list and P2 list Calculated.

[0038] Preferably, in step S6, the granularity of the vehicle speed state in the two-dimensional dynamic programming is 0.5. km / h The preferred SOC granularity is 0.25% SOC, which can be adjusted appropriately according to computing resources and accuracy requirements.

[0039] The beneficial effects of this invention are as follows:

[0040] The energy management method for hybrid electric vehicles designed in this invention, based on adaptive weighted dynamic programming, adapts to slope conditions and takes into account multiple performance objectives. Through an adaptive weight module, the weights of the objective function are dynamically adjusted based on the slope ahead and the state of charge (SOC). This enhances power response when going uphill, strengthens energy recovery when going downhill, and optimizes fuel economy on flat roads, thus solving the problem of single-objective optimization caused by fixed weights in traditional dynamic programming.

[0041] The present invention designs an adaptive weighted dynamic programming hybrid vehicle train energy management method, which improves overall performance through collaborative planning. It incorporates vehicle speed planning and battery energy allocation into the same dynamic programming model to achieve coupled optimization of the two, avoids "conflict between vehicle speed demand and energy supply" (such as the lag in energy allocation during rapid acceleration in traditional strategies), and improves driving smoothness and power response consistency. Attached Figure Description

[0042] Figure 1 The overall architecture flowchart of the hybrid vehicle train energy management method based on adaptive weighted dynamic programming.

[0043] Figure 2 This is a schematic diagram of the SOC feasible region in an embodiment of the present invention.

[0044] Figure 3 This is a schematic diagram of the speed-SOC planning results in an embodiment of the present invention.

[0045] Figure 4 This is a schematic diagram of the elevation of the road ahead in an embodiment of the present invention. Detailed Implementation

[0046] The present invention will be further described below with reference to the accompanying drawings and reference numerals.

[0047] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0048] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0049] like Figure 1 As shown, an adaptive weighted dynamic programming method for energy management of hybrid electric vehicles includes the following steps:

[0050] Step S1, Road Network Reconstruction: Utilize the vehicle network service system's terminal and map to obtain the road network ahead of the vehicle (6 km ahead). km Road gradient information, based on every 300 meters... mThe average slope sequence is calculated and used as input for subsequent steps.

[0051] Step S2, Feasible Domain Calculation: Based on the road gradient ahead, the driver's set cruise speed, and the acceptable speed fluctuation threshold, the planned speed feasible domain and SOC feasible domain are calculated using vehicle dynamics formulas and battery charging and discharging formulas. Based on the maximum engine torque, motor positive torque, and maximum motor negative torque of the vehicle's power system, the maximum and minimum accelerations achieved by the vehicle in each road segment are calculated. Starting from the cruise speed at the planning starting point, the calculation is iteratively performed to obtain the speed achievable domain, and then iteratively obtains the feasible domains of all road segments ahead.

[0052] The applied vehicle longitudinal dynamics formula is:

[0053] ;

[0054] in, T tq To output torque to the vehicle's powertrain. η T For the mechanical efficiency of the transmission system. i g This refers to the gear ratio of the transmission. i 0 Main reducer transmission ratio, f The rolling resistance coefficient, C D The air drag coefficient, A For windward area, α For road slope, u α For driving speed, m For the total load capacity of the vehicle, a To accelerate the vehicle, G To accelerate the vehicle from a standstill to 100 km / h The ratio of the average acceleration to the gravitational acceleration during the process.

[0055] For battery SOC, during forward calculation, the decrease in battery SOC is calculated based on the energy consumed when the motor outputs the maximum positive torque in each road segment, and the increase in battery SOC is calculated based on the energy consumed when the motor outputs the maximum negative torque in each road segment, thereby constructing the forward feasible region of SOC.

[0056] Because this method presupposes that the battery SOC remains the same as the initial value at the end of the planning process to ensure system sustainability, the feasible region of SOC must also converge to the initial value at the end. Therefore, it is necessary to back-calculate the feasible region of SOC from the end. During the back-calculation process, the engine is temporarily not involved in torque output. For the discharge process, the SOC change is always calculated based on the maximum positive motor torque until the SOC reaches the upper threshold. During this process, the speed change before and after is calculated simultaneously. If the speed reaches the upper speed limit first at the moment before the back-calculation, the positive torque output is paused in this stage, and the calculation directly proceeds to the previous stage. For the charging process, the SOC change is always calculated based on the maximum negative motor torque until the SOC reaches the lower threshold. During this process, the speed change before and after is calculated simultaneously. If the speed reaches the lower speed limit first at the moment before the back-calculation, the negative torque output is paused in this stage, and the calculation directly proceeds to the previous stage. After the back-calculation process is completed, the intersection of the forward and reverse feasible regions of SOC is taken to obtain the final feasible region.

[0057] The battery charge / discharge formula for battery SOC applications is as follows:

[0058] ;

[0059] in, energy down This represents the total power of the motor under discharge conditions. energy up This refers to the total power of the motor under charging conditions. eff m For motor efficiency, eff bat For battery efficiency, U cell Rated voltage, Ebat This represents the total energy of the battery.

[0060] Step S3: Establishing the pre-planning cost function model: In this step, only the vehicle speed is determined as the state variable of the dynamic programming algorithm. Considering four aspects, fuel quantity, driving comfort, deviation from the driver's cruise speed setting, and timeliness, corresponding weights are assigned to establish the pre-planning cost function model.

[0061] The formula for the pre-planning cost function model is as follows:

[0062] ;

[0063] in, f k+1 For the first k+1 The overall cost value, Q k+1 For the first k+1 The fuel consumption cost per unit, V k For the first k The vehicle speed, Vk+1 For the first k+1 The vehicle speed, For the first k+1 The and the first k The average speed of each vehicle V CC Set the speed for cruise control. t k+1 For the first k+1 The running time of each vehicle r 1 For pre-planned fuel quantity weighting, r 2 Weighting of pre-planned vehicle speed changes, r 3 To set speed deviation weights for pre-planning and cruise control, r 4 This is a weight for the timeliness of pre-planning.

[0064] Step S4, One-Dimensional Dynamic Programming of Vehicle Speed: This stage only involves planning within the vehicle speed range. The output provides adaptive optimization information for subsequent two-dimensional planning. The DP algorithm optimizes by iteratively calculating the state transition equation within the feasible speed region in step S2. The specific state transition equation is as follows:

[0065] ;

[0066] in, For the first k+1 The cost value per vehicle speed From u k arrive u k+1 The cost value matrix, For the first k The cost value per vehicle speed .

[0067] The speed trajectory obtained from the planning is denoted as V pre .

[0068] Step S5, Adaptive Weight and Cost Function Update: This step updates the expected vehicle speed offset cost function based on the speed trajectory information planned in step S4, and adaptively updates the weights of vehicle speed change and cruise setting vehicle speed offset based on the road gradient information ahead.

[0069] Considering the costs of speed variation and cruise speed deviation, and combining the speed planning results obtained in step S4... V pre The weights of the costs associated with speed changes and deviations from cruising speed are adaptively adjusted.

[0070] (1) For the cost of speed variation, the speed planning results will be used. V pre Differential calculations are performed to obtain the speed change sequence for each road segment. The absolute value of this sequence is then taken and normalized to obtain the adaptive reference coefficient for each road segment. P1 list ;

[0071] (2) For the cost of deviation from cruising speed, the speed planning results will be used. V pre Subtract cruise control speed V CC The speed deviation sequence of each road segment is obtained, and the absolute value and normalization operations of this sequence are performed to obtain the adaptive reference coefficient of each road segment. P2 list .

[0072] The formula for establishing the cost function model is as follows:

[0073] ;

[0074] in, f k+1 For the first k+1 The cost value per vehicle speed Q k+1 For the first k+1 Fuel consumption at a given vehicle speed V k For the first k The vehicle speed, V k+1 For the first k+1 The vehicle speed, For the first k+1 The and the first k The average speed of each vehicle The final result obtained in step S4 k+1 A planned speed trajectory, t k+1 For the first k+1 The running time of each vehicle R 1 As a weighted average of fuel quantity, R 2 As the weight of the change in vehicle speed, R 3 To set the speed deviation weight with cruise control, R 4 Timeliness is a weighting factor.

[0075] Taking into account both economic efficiency and transportation efficiency, the adaptive weighting function is constructed as follows:

[0076] ;

[0077] in, R 1 , R 4 Initial weighting coefficients r 1 , r 4 constant, R 2 , R 3 Optimize coefficients using adaptive weights P1 list and P2 list Calculated.

[0078] Step S6, Vehicle Speed ​​and SOC Dynamic Programming: The two-dimensional DP algorithm optimizes by iteratively calculating the state transition equation within the feasible region in step S2. The specific state transition equation is as follows:

[0079] ;

[0080] in, For the first k+1 (vehicle speed - battery) SOC The cost value of ) From u k arrive u k+1 The cost value matrix, For the first k (vehicle speed - battery) SOC The cost value of ) .

[0081] Step S7, Planning Vehicle Speed ​​and SOC Execution: The optimal vehicle speed sequence and SOC sequence determined in step S6 are output to the execution end for predictive energy management of the hybrid electric vehicle train; the motor output torque is determined by the planned SOC sequence, the actual required torque is subtracted from the motor output torque to obtain the engine torque, and the engine torque is controlled by a PID controller to achieve the planned target speed tracking and perform optimal energy efficiency cruise control.

[0082] Specific embodiments of the present invention

[0083] like Figure 1 As shown, an adaptive weighted dynamic programming method for hybrid vehicle train energy management includes the following steps:

[0084] The vehicle-to-everything (V2X) service system terminal and map box obtain the information in front of the vehicle at 6... km The road gradient information is sent to the VCU controller via the CAN bus using CAN signals. The VCU then responds in increments of 300 meters. mCalculate the average slope sequence to reconstruct the road network. After the calculation is completed, mark the effective map marker position as 1.

[0085] The VCU calculates the feasible speed and SOC (State of Charge) regions based on information such as the road gradient ahead, the driver's set cruise speed, and acceptable speed fluctuation thresholds, using vehicle dynamics formulas. It then calculates the maximum and minimum acceleration achievable by the vehicle on each road segment based on the sum of the engine's maximum torque, the motor's maximum positive torque, and the motor's maximum negative torque, further deriving the achievable speed region. This process is iteratively repeated to obtain the feasible regions for all road segments ahead. For the battery SOC feasible region, the corresponding SOC change is calculated based on the motor's maximum positive and negative torques. Forward and backward derivations are performed sequentially, and their intersection is taken to obtain the final SOC feasible region.

[0086] SOC feasible domain examples Figure 2 As shown: the purple part is the final SOC feasible region, the red part is the forward SOC feasible region, and the blue part is the reverse SOC feasible region.

[0087] After obtaining the feasible region, the VCU uses a one-dimensional dynamic programming algorithm with fixed weight parameters to calculate the planned speed sequence. Then, it uses this speed sequence information to optimize the adaptive weight function: for cost function optimization, the obtained speed planning sequence is used as the tracking target, replacing the target speed for speed deviation cost. For weight parameter optimization, the obtained speed sequence is subjected to differencing and subtraction of the cruising speed, absolute value taking, and normalization operations. The obtained data is then used to proportionally reduce the weights for speed change cost and speed deviation cost, thereby ensuring that the constraint of the cost term is relaxed in road sections requiring speed adjustment. The code for this part is as follows:

[0088]

[0089] in, P1 list and P2 list These represent the information sequences obtained by differentiating the original velocity planning sequence and normalizing it after subtracting the cruise speed, respectively. The VCU then applies a two-dimensional dynamic programming algorithm to calculate the planned velocity and SOC state trajectory, obtaining the final planning result, as shown below. Figure 3 As shown, blue represents the velocity planning result, and red represents the SOC planning result, as... Figure 4 The image shows the elevation of the road ahead, demonstrating how energy optimization is achieved during downhill deceleration.

[0090] Finally, the VCU combines the planned SOC trajectory with the speed information calculated from the speed trajectory to derive the required motor torque for each stage. The motor torque is then sent to the HCU for torque control deployment. For speed planning, the VCU applies PID control to control the engine torque, achieving planned speed tracking.

[0091] This invention is not limited to the above-described optional embodiments. Anyone can derive other various forms of products under the guidance of this invention. However, regardless of any changes made in their shape or structure, any technical solution that falls within the scope of the claims of this invention shall be protected by this invention.

Claims

1. An adaptive weighted dynamic programming method for energy management of hybrid electric vehicles, characterized in that, Includes the following steps: S1. Road network reconstruction: Obtain the slope information of the road ahead using the terminal and map of the vehicle network service system; S2. Feasible Domain Calculation: Based on the road gradient ahead, the driver's set cruise speed, and the acceptable speed fluctuation threshold, the planned speed feasible domain and SOC feasible domain are calculated in combination with vehicle dynamics formulas and battery charging and discharging formulas. S3. Establishment of the pre-planning cost function model: Determine the vehicle speed as the state variable of the dynamic programming algorithm and establish the pre-planning cost function model; S4. One-dimensional dynamic programming of vehicle speed: Planning within the vehicle speed range, the output provides adaptive optimization information for subsequent two-dimensional planning. The DP algorithm optimization utilizes the state transition equation to iteratively calculate within the feasible speed region in step S2. The planned speed trajectory is denoted as... V pre ; S5. Adaptive Weight and Cost Function Update: Based on the pre-planned speed trajectory obtained in step S4, generate speed change sequences and cruise deviation sequences, and calculate reference coefficients accordingly. These coefficients are used to proportionally update the speed change weights and cruise deviation weights in the pre-planned penalty function. R2, R3 At the same time, the target for deviation will be changed from the cruise control set speed to... V pre This results in the updated penalty function; S6. Dynamic programming of vehicle speed and SOC: Using vehicle speed and SOC as state variables, under the constraints of the feasible region of vehicle speed and the feasible region of SOC constructed in step S2, two-dimensional dynamic programming is used to find the optimal result by using the updated penalty function to obtain the final vehicle speed sequence and the optimal SOC sequence for subsequent execution control. S7. Planning vehicle speed and SOC execution: The optimal vehicle speed sequence and SOC sequence determined in step S6 are output to the execution end for predictive energy management of the hybrid electric vehicle train; the motor output torque is determined by the planned SOC sequence, the actual required torque is subtracted from the motor output torque to obtain the engine torque, and the engine torque is controlled by a PID controller to achieve the planned target speed tracking and perform optimal energy efficiency cruise control.

2. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 1, characterized in that, In step S1, the slope information is 6 meters ahead of the vehicle. km Road gradient information, based on every 300 meters... m The average slope sequence within the range is calculated and used as input for subsequent steps.

3. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 1, characterized in that, The vehicle dynamics formula in step S2 is: ； in, T tq To output torque to the vehicle's powertrain. η T For the mechanical efficiency of the transmission system. i g This refers to the gear ratio of the transmission. i 0 Main reducer transmission ratio, f The rolling resistance coefficient, C D The air drag coefficient, A For windward area, α For road slope, u α For driving speed, m For the total load capacity of the vehicle, a To accelerate the vehicle, G To accelerate the vehicle from a standstill to 100 km / h The ratio of the average acceleration to the gravitational acceleration during the process.

4. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 1, characterized in that, The construction rule for the planned vehicle speed feasible domain in step S2 is as follows: based on the maximum engine torque, motor positive torque and maximum motor negative torque of the vehicle power system, calculate the maximum acceleration and minimum acceleration reached by the vehicle in each road segment, start the iterative calculation from the cruising speed at the planning starting point to obtain the vehicle speed feasible domain, and iteratively obtain the vehicle speed feasible domain for all road segments ahead. The construction rules for the SOC feasible region in step S2 are as follows: During forward calculation, the reduction in battery SOC is calculated based on the energy consumed when the motor outputs the maximum positive torque in each road segment, and the increase in battery SOC is calculated based on the energy replenished when the motor outputs the maximum negative torque in each road segment, thereby constructing the forward feasible region of SOC. The reverse derivation of the SOC feasible region is as follows: with the constraint that the endpoint SOC remains consistent with the initial SOC, the maximum positive / negative motor torque is used as input, and the SOC feasible interval is deduced from the endpoint to the starting point in reverse. During the process, the engine does not participate in torque output, resulting in the reverse feasible region of SOC. Finally, the intersection of the forward feasible region of SOC and the reverse feasible region of SOC is taken to obtain the final SOC feasible region.

5. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 4, characterized in that, The battery charge / discharge formula for the battery's state of charge (SOC) is as follows: ； in, energy down This represents the total power of the motor under discharge conditions. energy up This refers to the total power of the motor under charging conditions. eff m For motor efficiency, eff bat For battery efficiency, U cell Rated voltage, Ebat This represents the total energy of the battery.

6. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 1, characterized in that, In step S3, vehicle speed is set as the state variable of the dynamic programming algorithm. Simultaneously, the pre-planning cost function includes four cost items: fuel consumption, driving comfort, cruise speed deviation, and driving timeliness. Each cost item is assigned a corresponding initial weight. r 1 ~r 4 The formula for the pre-planning cost function model is as follows: ； in, f k+1 For the first k+1 The overall cost value, Q k+1 For the first k+1 The fuel consumption cost per unit, V k For the first k The vehicle speed, V k+1 For the first k+1 The vehicle speed, For the first k+1 The and the first k The average speed of each vehicle V CC Set the speed for cruise control. t k+1 For the first k+1 The running time of each vehicle r 1 For pre-planned fuel quantity weighting, r 2 Weighting of pre-planned vehicle speed changes, r 3 To set speed deviation weights for pre-planning and cruise control, r 4 This is a weight for the timeliness of pre-planning.

7. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 6, characterized in that, The same state transition equation is used in both steps S4 and S6, specifically: ； The variable definitions defined in step S4 include: For the first k+1 The cost value per vehicle speed From u k arrive u k+1 The cost value matrix, For the first k The cost value per vehicle speed ; The variable definitions defined in step S6 include: For the first k+1 The cost value of each two-dimensional state, that is, the cost value of the two state variables: vehicle speed and battery SOC. To be from two-dimensional state u k To two-dimensional state u k+1 The cost value matrix, For the first k The cost value of a two-dimensional state. .

8. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 7, characterized in that, The cost function model formula in step S5 is as follows: ； in, f k+1 For the first k+1 The cost value per vehicle speed Q k+1 For the first k+1 Fuel consumption at a given vehicle speed V k For the first k The vehicle speed, V k+1 For the first k+1 The vehicle speed, For the first k+1 The and the first k The average speed of each vehicle The final result obtained in step S4 k+1 A planned speed trajectory, t k+1 For the first k+1 The running time at each vehicle speed R 1 As a weighted average of fuel quantity, R 2 As the weight of the change in vehicle speed, R 3 To set the speed deviation weight with cruise control, R 4 Timeliness is a weighting factor.

9. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 8, characterized in that, The specific steps for adaptively updating the weights of vehicle speed change and cruise set speed offset based on the road gradient information in step S5 are as follows: (1) For the cost of speed variation, the speed planning results will be used. V pre Differential calculations are performed to obtain the speed change sequence for each road segment. The absolute value of this sequence is then taken and normalized to obtain the adaptive reference coefficient for each road segment. P1 list ; (2) For the cost of deviation from cruising speed, the speed planning results will be used. V pre Subtract cruise control speed V CC The speed deviation sequence of each road segment is obtained, and the absolute value and normalization operations of this sequence are performed to obtain the adaptive reference coefficient of each road segment. P2 list .

10. The hybrid vehicle train energy management method based on adaptive weighted dynamic programming according to claim 9, characterized in that, The adaptive weight function in step S5 is constructed as follows: ； in, R 1 , R 4 The initial weighting coefficient is r1. r 4 constant, R 2 , R 3 Optimize coefficients using adaptive weights P1 list and P2 list Calculated.

Images