Vehicle oil and electricity distribution method and electronic equipment

Abstract

The invention provides a vehicle oil and electricity distribution method and electronic equipment. The method comprises the steps that current driving state information of a vehicle is acquired; determining a current driving scene of the vehicle according to the current driving state information of the vehicle; according to the current driving scene of the vehicle, determining a target oil-electricity distribution model corresponding to the current driving scene of the vehicle; wherein the oil-electricity distribution model comprises scene oil-electricity distribution models and a general oil-electricity distribution model, and each scene oil-electricity distribution model is obtained through training of the corresponding driving scene data set and the general oil-electricity distribution model; the universal oil-electricity distribution model is obtained by performing reinforcement learning on a plurality of training samples based on a multi-target reward function and physical constraint conditions, and the physical constraint conditions are used for constraining and limiting different battery charge states and an interval range where an engine power ratio is located under different engine rotating speeds; and inputting the current driving state information of the vehicle into the target oil-electricity distribution model to obtain a target engine power ratio.

Description

Technical Field

[0001] This application relates to the field of vehicle energy management technology, and more specifically, to a vehicle fuel-electricity distribution method and electronic device. Background Technology

[0002] Traditional range-extended electric vehicles primarily employ rule-based control strategies (such as thermostat strategies and power-following strategies) for energy distribution control. This rule-based approach achieves fuel-electricity distribution by setting preset logical thresholds (such as starting the engine when the battery SOC is below a certain value). While its logic is simple and computationally efficient, it lacks flexibility and struggles to adapt to complex road conditions and changes in driving style. Summary of the Invention

[0003] The purpose of this application is to provide a vehicle fuel-electricity distribution method and electronic device to solve the problem that the current rule-based control strategy energy distribution control method has poor flexibility and is difficult to adapt to complex road conditions and changes in driving style.

[0004] Firstly, this application provides a vehicle fuel-electricity distribution method, which includes: acquiring the vehicle's current driving state information; wherein the driving state information includes the vehicle state, driving environment state, and driver state; determining the vehicle's current driving scenario based on the vehicle's current driving state information; determining the target fuel-electricity distribution model corresponding to the vehicle's current driving scenario based on the vehicle's current driving scenario; wherein the fuel-electricity distribution model includes a scenario-based fuel-electricity distribution model and a general fuel-electricity distribution model, each scenario-based fuel-electricity distribution model is trained using a corresponding driving scenario dataset and a general fuel-electricity distribution model, the general fuel-electricity distribution model is obtained by reinforcement learning on multiple training samples based on a multi-objective reward function and physical constraints, each training sample includes the vehicle's historical driving state information, and the physical constraints are used to constrain and limit the range of engine power ratios under different battery states of charge and different engine speeds; and inputting the vehicle's current driving state information into the target fuel-electricity distribution model to obtain the target engine power ratio output by the target fuel-electricity distribution model.

[0005] This solution provides a vehicle fuel-electric distribution method that determines the vehicle's driving scenario by integrating multi-dimensional information such as vehicle status, driving environment status, and driver status. Based on different driving scenarios, corresponding fuel-electric distribution models are pre-assigned, allowing the actual driving scenario to be matched with the corresponding target fuel-electric distribution model. Then, based on the target fuel-electric distribution model, fuel-electric distribution is achieved for that driving scenario, outputting the target engine power ratio. This achieves scenario-based, precise fuel-electric distribution, improving efficient decision-making and rapid adaptation under complex road conditions. Furthermore, this solution's fuel-electric distribution model incorporates physical constraints to ensure that the battery and engine operate within a safe and efficient range, improving vehicle reliability and economy.

[0006] In an optional implementation of the first aspect, determining the target fuel-electric distribution model corresponding to the current driving scenario of the vehicle includes: determining whether the current driving scenario of the vehicle belongs to a target preset scenario; wherein the target preset scenario is any preset driving scenario among multiple preset driving scenarios; if it is determined that the current driving scenario of the vehicle belongs to the target preset driving scenario, then searching for the target scenario fuel-electric distribution model corresponding to the target driving scenario; wherein each scenario fuel-electric distribution model corresponds to a preset driving scenario; and determining the target scenario fuel-electric distribution model as the target fuel-electric distribution model corresponding to the current driving scenario of the vehicle.

[0007] In the above implementation method, this solution clearly defines the selection logic of the scenario-based fuel-electricity distribution model under the preset scenario, ensuring the pertinence and optimization of the fuel-electricity distribution strategy under specific scenarios, and further improving fuel economy and power performance.

[0008] In an optional implementation of the first aspect, the method further includes: if it is determined that the current driving scenario of the vehicle does not belong to the target preset driving scenario, then the general oil-electricity distribution model is determined as the target oil-electricity distribution model corresponding to the current driving scenario of the vehicle.

[0009] In the above implementation method, this solution supplements the selection rules of the general model in non-preset scenarios, ensuring that there is an appropriate oil-electricity distribution scheme in all driving scenarios, thereby enhancing the versatility and robustness of the method.

[0010] In an optional embodiment of the first aspect, after obtaining the target engine power ratio output by the target fuel-electric distribution model, the method further includes: using a model predictive control algorithm to generate a target engine power curve based on the target engine power ratio; controlling the vehicle to drive according to the target engine power curve; collecting engine data and battery data of the vehicle in real time after a preset time period; obtaining engine prediction data and battery prediction data obtained by multi-objective reward function calculation in the target fuel-electric distribution model; calculating a deviation value based on the engine data, battery data, engine prediction data, and battery prediction data; wherein the deviation value is used to adjust and correct the target fuel-electric distribution model.

[0011] In the above implementation method, this solution achieves real-time optimization of the oil-electricity allocation model through model predictive control and deviation correction mechanism, improves the model prediction accuracy and control effect, and enables long-term sustainable optimization of the oil-electricity allocation strategy.

[0012] In an optional implementation of the first aspect, before obtaining the current driving state information of the vehicle, the method further includes: obtaining training samples; wherein the training samples include historical driving state information of the vehicle, which includes historical vehicle state, historical driving environment state, and historical driver state; employing a proximal policy optimization model, utilizing physical constraints and a multi-objective reward function to perform reinforcement learning on the training samples to obtain a general fuel-electricity allocation model; extracting shared features under different scenarios in the general fuel-electricity allocation model; obtaining driving scenario datasets for various pre-classified driving scenarios; and training a scenario fuel-electricity allocation model corresponding to each driving scenario based on the shared features under different scenarios in the general fuel-electricity allocation model and the driving scenario datasets for each driving scenario.

[0013] In the above implementation method, this solution constructs a hierarchical training architecture of a general model and a scenario model. The general model ensures basic adaptability, while the scenario model is optimized for specific scenarios, taking into account both the generality of the model and the specificity of the scenario. The scenario model is trained by using shared features, which reduces training costs and improves the convergence speed of the model.

[0014] In an optional implementation of the first aspect, a proximal policy optimization model is employed, utilizing physical constraints and a multi-objective reward function to perform reinforcement learning on the training samples to obtain a general fuel-electricity allocation model. This includes: extracting a high-dimensional state vector for each feature in the training samples to obtain a full-dimensional state vector corresponding to the training samples; determining the initial engine power ratio probability distribution of the vehicle in all-dimensional states using the proximal policy optimization model based on the full-dimensional state vector corresponding to the training samples; constraining the initial engine power ratio probability distribution using physical constraints to obtain a constrained initial engine power ratio probability distribution; calculating the reward / penalty signal corresponding to the constrained initial engine power ratio probability distribution based on preset evaluation conditions using a multi-objective reward function; and optimizing and adjusting the network parameters of the proximal policy optimization model based on the reward / penalty signal until the optimized and adjusted network parameters converge to obtain the general fuel-electricity allocation model.

[0015] In the above implementation method, this scheme uses a proximal policy optimization model combined with physical constraints and a multi-objective reward function to train a general model. The proximal policy optimization model improves the stability and convergence speed of reinforcement learning, while physical constraints and a multi-objective reward function ensure that the model output conforms to the physical characteristics of the vehicle and takes into account multi-dimensional performance objectives.

[0016] In an optional implementation of the first aspect, the physical constraints include: limiting the proportion of engine power of the vehicle using a first constraint based on the current SOC value of the battery; and limiting the proportion of engine power of the vehicle using a second constraint based on the real-time speed of the engine.

[0017] The above implementation method sets physical constraints from two dimensions: battery SOC and engine speed, to avoid battery overcharging and over-discharging and inefficient engine operation, thereby extending the service life of the battery and engine and improving the overall reliability of the vehicle.

[0018] In an optional implementation of the first aspect, the first limiting condition includes: when the battery SOC is less than a first SOC threshold, limiting the engine power percentage to be greater than a first power percentage threshold; when the battery SOC is greater than a second SOC threshold, limiting the engine power percentage to be less than a second power percentage threshold; wherein the first SOC threshold is less than the second SOC threshold, and the first power percentage threshold is greater than the second power percentage threshold.

[0019] In the above implementation method, this solution refines the battery SOC constraints, accurately controls the engine power ratio based on the SOC threshold, effectively maintains the battery's health, and avoids battery damage.

[0020] In an optional implementation of the first aspect, the second limiting condition is: limiting the proportion of engine power of the vehicle within the calibrated thermal efficiency range corresponding to the real-time speed of the engine; wherein each speed corresponds to a thermal efficiency range.

[0021] In the above implementation method, this solution ensures that the engine always operates in the high-efficiency range by matching the engine speed with the thermal efficiency range, thereby reducing fuel consumption and emissions and improving fuel economy.

[0022] Secondly, this application provides a vehicle fuel-electric distribution device, which includes an acquisition module, a determination module, and an input module. The acquisition module is used to acquire the current driving state information of the vehicle, wherein the driving state information includes the vehicle state, the driving environment state, and the driver state. The determination module is used to determine the target fuel-electric distribution model corresponding to the current driving scenario of the vehicle. The fuel-electric distribution model includes a scenario-based fuel-electric distribution model and a general fuel-electric distribution model. Each scenario-based fuel-electric distribution model is obtained by training on the corresponding driving scenario dataset and the general fuel-electric distribution model. The general fuel-electric distribution model is obtained by reinforcement learning on multiple training samples based on a multi-objective reward function and physical constraints. Each training sample includes the vehicle's historical driving state information. The physical constraints are used to constrain and limit the range of the engine power ratio under different battery states of charge and different engine speeds. The input module is used to input the current driving state information of the vehicle into the target fuel-electric distribution model to obtain the target engine power ratio output by the target fuel-electric distribution model.

[0023] The vehicle fuel-electric distribution device designed above determines the vehicle's driving scenario by integrating multi-dimensional information such as vehicle status, driving environment status, and driver status. It pre-allocates corresponding fuel-electric distribution models based on different driving scenarios, enabling the matching of the target fuel-electric distribution model to the actual determined driving scenario. Then, based on the target fuel-electric distribution model, it achieves fuel-electric distribution under that driving scenario, outputting the target engine power ratio. This achieves scenario-based precise fuel-electric distribution, improving efficient decision-making and rapid adaptation under complex road conditions. Furthermore, the fuel-electric distribution model of this solution incorporates physical constraints to ensure that the battery and engine operate within a safe and efficient range, improving vehicle reliability and economy.

[0024] Thirdly, this application provides an electronic device including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the method described in the first aspect and any optional embodiment of the first aspect.

[0025] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, performs the method described in the first aspect or any optional embodiment of the first aspect.

[0026] Fifthly, this application provides a computer program product, including a computer program / instructions, which, when executed by a processor, perform the methods described in the first aspect and any optional implementation thereof.

[0027] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0028] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments of this application will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a first flowchart illustrating the vehicle fuel-electricity distribution method provided in an embodiment of this application; Figure 2 This is a second flowchart illustrating the vehicle fuel-electricity distribution method provided in an embodiment of this application. Figure 3 A schematic diagram of the third process of the vehicle fuel-electricity distribution method provided in the embodiments of this application; Figure 4 This is a schematic diagram of the structure of the vehicle fuel-electricity distribution device provided in the embodiments of this application; Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0030] Icons: 400 - Acquisition module; 410 - Determining module; 420 - Input module; 5 - Electronic device; 501 - Processor; 502 - Memory; 503 - Communication bus. Detailed Implementation

[0031] The embodiments of the technical solution of this application will now be described in detail with reference to the accompanying drawings. These embodiments are only used to more clearly illustrate the technical solution of this application and are therefore merely examples, and should not be used to limit the scope of protection of this application.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the application; the terms “comprising” and “having”, and any variations thereof, in the specification, claims, and foregoing description of the drawings are intended to cover non-exclusive inclusion.

[0033] In the description of the embodiments of this application, technical terms such as "first" and "second" are used only to distinguish different objects and should not be construed as indicating or implying relative importance or implicitly specifying the number, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, "multiple" means two or more, unless otherwise explicitly defined.

[0034] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0035] In the description of the embodiments in this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0036] In the description of the embodiments of this application, the term "multiple" refers to two or more (including two), similarly, "multiple sets" refers to two or more (including two sets), and "multiple pieces" refers to two or more (including two pieces).

[0037] In the description of the embodiments of this application, the technical terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing the embodiments of this application and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0038] In the description of the embodiments of this application, unless otherwise expressly specified and limited, technical terms such as "installation," "connection," "joining," and "fixing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. For those skilled in the art, the specific meaning of the above terms in the embodiments of this application can be understood according to the specific circumstances.

[0039] Traditional range-extended electric vehicles primarily employ rule-based control strategies (such as thermostat strategies and power-following strategies) for energy distribution control. This rule-based approach achieves fuel-electricity distribution by setting preset logical thresholds (such as starting the engine when the battery SOC is below a certain value). While its logic is simple and computationally efficient, it lacks flexibility and struggles to adapt to complex road conditions and changes in driving style.

[0040] To address the aforementioned issues, this application designs a vehicle fuel-electric distribution method and electronic device. By integrating multi-dimensional information such as vehicle status, driving environment status, and driver status, the method determines the vehicle's driving scenario and pre-allocates corresponding fuel-electric distribution models based on different driving scenarios. This allows the method to match the actual driving scenario with the corresponding target fuel-electric distribution model, and then achieve fuel-electric distribution under that driving scenario, outputting the target engine power ratio. This enables scenario-based, precise fuel-electric distribution, improving efficient decision-making and rapid adaptation under complex road conditions. Furthermore, the fuel-electric distribution model in this solution incorporates physical constraints to ensure that the battery and engine operate within a safe and efficient range, enhancing vehicle reliability and economy. In addition, this solution quickly invokes the corresponding scenario-based fuel-electric distribution model for fuel-electric distribution in clearly defined preset scenarios, thereby making rapid decisions and meeting the stringent real-time requirements of the vehicle system. In non-preset scenarios, a general fuel-electric distribution model is selected for fuel-electric distribution, ensuring that there is an appropriate fuel-electric distribution scheme for all driving scenarios. This balances the millisecond-level response efficiency of common scenarios with the fine-grained optimization accuracy of complex scenarios, achieving the hierarchical control objective of "rapid decision-making in simple scenarios and in-depth optimization in complex scenarios," and enhancing the versatility and robustness of the method.

[0041] Based on the above ideas, this application first provides a vehicle fuel-electricity distribution method. This method can be applied to computing devices, including but not limited to cloud servers, vehicle controllers, in-vehicle terminals, and vehicle-mounted computers, etc. Figure 1 As shown, the vehicle's fuel-electricity distribution method can be implemented in the following ways: Step S100: Obtain the current driving status information of the vehicle.

[0042] Step S110: Determine the current driving scenario of the vehicle based on the vehicle's current driving status information.

[0043] Step S120: Determine the target fuel-electricity distribution model corresponding to the current driving scenario of the vehicle.

[0044] Step S130: Input the vehicle's current driving status information into the target fuel-electricity distribution model to obtain the target engine power ratio output by the target fuel-electricity distribution model.

[0045] In the above-described implementation, this solution can acquire the vehicle's current driving status information. This current driving status information may include multi-dimensional information, such as vehicle status, driving environment status, and driver status. The vehicle status represents the vehicle's current internal state, such as battery SOC, engine speed, motor power demand, and battery temperature. These vehicle states can be collected by various sensors and BMS installed in the vehicle. The driving environment status represents the external environmental data of the vehicle's current driving, such as real-time traffic congestion level, road gradient, weather, distance to vehicles ahead, and speed prediction for a future preset mileage. The driver status represents the current driver's driving style, such as accelerator pedal opening change rate and historical driving style classification, which can be collected by an onboard camera and pedal sensor. In addition to the above-mentioned status information, the current driving status information may also include the grid interaction status, which may include real-time electricity price (yuan / kWh), peak and valley time period identifiers (0=valley electricity, 1=peak electricity), battery capacity that can participate in V2G (kWh), etc., which can be obtained through information interaction with the grid.

[0046] Having obtained the vehicle's current driving status information through the aforementioned methods, this solution can determine the vehicle's current driving scenario based on this information. Specifically, this solution can identify the driving scenario based on the vehicle's status, the driving environment's status, and the driver's status, thereby obtaining the vehicle's current driving scenario. For the identification of the driving scenario, this solution can employ a clustering model or clustering algorithm.

[0047] When the current driving scenario of the vehicle is identified through the above method, this solution can determine the target fuel-electric distribution model corresponding to the current driving scenario. The fuel-electric distribution model includes a scenario-based fuel-electric distribution model and a general fuel-electric distribution model. Each scenario-based fuel-electric distribution model is trained using the corresponding driving scenario dataset and the general fuel-electric distribution model. The general fuel-electric distribution model is obtained by reinforcement learning on multiple training samples based on a multi-objective reward function and physical constraints. Each training sample includes the vehicle's historical driving state information. The physical constraints are used to constrain and limit the range of the engine power ratio under different battery states of charge and different engine speeds.

[0048] Specifically, this solution can determine whether the vehicle's current driving scenario belongs to a target preset scenario, where the target preset scenario is any one of multiple preset driving scenarios. Specifically, this solution can predefine multiple driving scenarios (such as urban congestion, highway cruising, mountain climbing, etc.) using a clustering algorithm. In this case, this solution can use the clustering algorithm to identify whether the vehicle's current driving scenario belongs to any one of the multiple preset driving scenarios based on the vehicle's current driving status information. For example, it can determine whether the vehicle's current driving scenario is any one of the preset driving scenarios such as urban congestion, highway cruising, or mountain climbing.

[0049] If the above method determines that the vehicle's current driving scenario belongs to a target preset driving scenario, then this solution searches for the target scenario's fuel-electric distribution model based on the target driving scenario, and then determines the target scenario's fuel-electric distribution model as the target fuel-electric distribution model corresponding to the vehicle's current driving scenario. Each scenario's fuel-electric distribution model corresponds to a preset driving scenario.

[0050] If it is determined through the above method that the current driving scenario of the vehicle does not belong to the target preset driving scenario, then the general oil-electricity distribution model is determined as the target oil-electricity distribution model corresponding to the current driving scenario of the vehicle.

[0051] Based on the above method, this solution can determine the target fuel-electric distribution model corresponding to the current driving scenario of the vehicle. The current driving status information of the vehicle can be input into the target fuel-electric distribution model to obtain the target engine power ratio output by the target fuel-electric distribution model, and then the fuel-electric distribution control of the vehicle can be carried out using the target engine power ratio.

[0052] Specifically, this solution can convert the output target engine power ratio α value into an engine throttle opening signal and a battery DC / DC converter control signal, and transmit them to the ECU and BMS in real time via the CAN bus, thereby achieving dynamic distribution of oil and electric energy while ensuring vehicle power performance.

[0053] The vehicle fuel-electric distribution method described above determines the vehicle's driving scenario by integrating multi-dimensional information such as vehicle status, driving environment status, and driver status. It pre-allocates corresponding fuel-electric distribution models based on different driving scenarios, ensuring that the actual driving scenario can be matched with the corresponding target fuel-electric distribution model. Then, based on the target fuel-electric distribution model, the fuel-electric distribution under that driving scenario is achieved, outputting the target engine power ratio. This enables scenario-based, precise fuel-electric distribution, improving efficient decision-making and rapid adaptation under complex road conditions. Furthermore, the fuel-electric distribution model in this solution incorporates physical constraints to ensure that the battery and engine operate within a safe and efficient range, enhancing vehicle reliability and economy.

[0054] Specifically, this solution can obtain a general oil-electricity allocation model and a scenario-based oil-electricity allocation model through training in the following manner, such as... Figure 2 As shown, it includes: Step S200: Obtain training samples.

[0055] Step S210: The model is optimized using a proximal strategy. Physical constraints and a multi-objective reward function are used to perform reinforcement learning on the training samples to obtain a general oil-electricity allocation model.

[0056] Step S220: Extract shared features from different scenarios in the general oil-electricity allocation model.

[0057] Step S230: Obtain a pre-classified dataset of driving scenarios under various driving scenarios.

[0058] Step S240: Based on the shared features of different scenarios in the general oil-electricity allocation model and the driving scenario dataset for each driving scenario, train the scenario oil-electricity allocation model corresponding to each driving scenario.

[0059] In the above implementation method, the training samples include vehicle historical driving status information, which includes vehicle historical status, driving environment historical status, and driver historical status.

[0060] Specifically, this solution can collect full-domain data through the collaboration of vehicle-to-everything (V2X) and on-board sensors: using 5G / C-V2X modules to obtain real-time traffic conditions (congestion index, traffic light timing), weather forecasts (extreme weather warnings), and road network structure information in the cloud; and at the same time, collecting internal and external state data such as battery SOC, engine speed, road gradient, and distance to vehicles ahead through devices such as BMS, ECU, radar, and cameras to obtain training samples.

[0061] With training samples obtained through the above methods, this scheme employs a proximal policy optimization model. Utilizing physical constraints and a multi-objective reward function, reinforcement learning is applied to the training samples to obtain a general fuel-electricity allocation model. Specifically, this scheme extracts the high-dimensional state vector of each feature in the training samples to obtain the corresponding full-dimensional state vector. Then, using the proximal policy optimization model, based on the full-dimensional state vector of the training samples, the initial engine power ratio probability distribution of the vehicle under all-dimensional states is determined. Physical constraints are applied to constrain the initial engine power ratio probability distribution, resulting in a constrained initial engine power ratio probability distribution. A multi-objective reward function is used to calculate the reward / penalty signal corresponding to the constrained initial engine power ratio probability distribution based on preset evaluation conditions. The network parameters of the proximal policy optimization model are then optimized and adjusted based on the reward / penalty signal until the optimized network parameters converge to obtain the general fuel-electricity allocation model.

[0062] The physical constraints include: limiting the vehicle's engine power ratio based on the battery's current SOC value using a first constraint; and limiting the vehicle's engine power ratio based on the engine's real-time rotational speed using a second constraint. The first constraint includes: limiting the engine power ratio to be greater than a first power ratio threshold when the battery SOC is less than a first SOC threshold; and limiting the engine power ratio to be less than a second power ratio threshold when the battery SOC is greater than a second SOC threshold; wherein the first SOC threshold is less than the second SOC threshold, and the first power ratio threshold is greater than the second power ratio threshold. The second constraint is: limiting the vehicle's engine power ratio within a calibrated thermal efficiency range corresponding to the engine's real-time rotational speed; wherein each rotational speed corresponds to a thermal efficiency range.

[0063] Specifically, such as Figure 3 As shown, the above training process can be implemented in the following ways: First, multi-source data fusion processing generates a full-dimensional state vector: specifically, it integrates the vehicle's internal state (battery SOC, engine speed, motor power demand, battery temperature), external environmental data (real-time traffic congestion level, road slope, and predicted speed for the next 2km), driver intent (accelerator pedal opening change rate, historical driving style classification), grid interaction state (real-time electricity price (yuan / kWh), peak and off-peak time indicators (0=off-peak, 1=peak), and battery capacity for V2G participation (kWh); physical constraint parameters (power ratio range corresponding to the engine's lowest / highest efficiency (dynamically calculated based on real-time speed), and battery charging and discharging safety current threshold), forming a state space with 25+ dimensions. Then, it fuses multi-source sensor data through Kalman filtering, normalizes the numerical features using standardization (Z-score), and preserves temporal trend features (such as the slope of SOC change over the past minute), thereby obtaining a full-dimensional state vector to provide structured input for the model.

[0064] Then, the policy network defines the continuous action space and imposes physical constraints. Specifically, this scheme defines the engine power ratio as a continuous variable. This indicates the proportion of engine output power to the total power demand of the vehicle (e.g., =0.6 indicates that the engine provides 60% of the power and the battery provides 40%. The policy network outputs the probability distribution of consecutive actions through the Proximal Policy Optimization (PPO) algorithm. ,in: A 25+ dimension state vector (including SOC, speed, road conditions, etc.) is defined for S201. These are the policy network parameters (weights of a fully connected neural network).

[0065] Policy network output formula: ; ; in, The mean power percentage output of the mean network is the optimal power percentage. The standard deviation of the action distribution output by the standard deviation network; The exploration noise, which conforms to a standard normal distribution, is used to balance exploration and utilization.

[0066] Assuming a safe state of charge (SOC) range of [20%, 80%] is defined for the battery, the output is determined using a hard constraint truncation strategy: When SOC < 25% (low battery warning): The engine power ratio is required to be ≥40%, and priority is given to ensuring battery charging (to prevent the SOC from falling below the 20% safety limit).

[0067] When SOC > 75% (high charge protection): Limit engine power to ≤30% to reduce unnecessary power generation (avoid exceeding the 80% safety limit for SOC).

[0068] Based on the real-time engine speed n (rpm), query the pre-calibrated high-efficiency power ratio range. For example, when n=1800rpm, the high efficiency range is [0.3, 0.75] (corresponding to the operating condition with thermal efficiency >35%).

[0069] The constraint formula is: ; The clip(·) function is used to truncate the output, ensuring the final output... It falls within the high-efficiency range, and ≥90% of the decisions satisfy the engine thermal efficiency >35%.

[0070] We will continue to construct a weighted, adaptive multi-objective reward function to replace the fixed weighted summation, achieving a real-time balance between energy efficiency, economy, and safety. Specifically, we will construct a three-dimensional reward system encompassing energy efficiency, economy, and safety, with weights... ,、 ,、 The system is dynamically adjusted based on real-time operating conditions, replacing the traditional fixed weighting. ; Among them, the energy efficiency bonus is the core objective: ; This represents the actual energy consumption under the current action (kWh / 100km). The benchmark energy consumption under the same working conditions is obtained through historical data statistics, and the reward is inversely proportional to the energy consumption.

[0071] V2G scenario economic incentives: ; The real-time electricity price (RMB / kWh) is derived from the S201 grid interaction status. This represents the amount of electricity exchanged between the battery and the power grid (negative for charging, positive for discharging). This is an indicator function that encourages discharging (V2G revenue) only when the electricity price is above the average.

[0072] Security rewards are used to enforce compliance. ; To constrain the penalty coefficient for violations (e.g., when SOC < 20%), =10); Constraint violations include situations such as exceeding the SOC limit or the engine leaving the high-efficiency zone, and negative rewards will be given when triggered.

[0073] Dynamically adjust weights using a condition classifier: ; This is a scene feature vector (road conditions, SOC, etc. from S201); (·) represents a weight generation network (such as a lightweight neural network or a rule engine); For example, in urban traffic congestion scenarios: =0.6、 =0.2、 If the value is 0.2, then energy efficiency and battery protection should be prioritized. Off-peak electricity hours (S201 peak-valley indicator = 0): =0.5、 =0.3、 =0.2, then prioritize the economic benefits of low-cost charging.

[0074] In the above implementation process, the multi-objective reward function generates reward and penalty signals (positive rewards or negative penalties) by quantifying the decision-making effects of α in the dimensions of energy efficiency (such as the comparison between actual energy consumption and benchmark energy consumption), economy (such as V2G revenue under real-time electricity price), and safety (such as whether SOC is in the 25%-75% range). For example, a positive reward is given to decisions that have better energy efficiency than expected (energy consumption reduction > 5%) or SOC fluctuations controlled within the safe range (25%-75%), and a negative penalty is imposed on decisions that deviate from the constraints (such as SOC < 25% or engine load > 80%) or energy consumption exceeding the standard.

[0075] The model (PPO policy network) learns the mapping relationship between "state s and optimal α" based on the reward and punishment signal. After outputting the probability distribution of continuous actions, it prioritizes the α that maximizes the reward function value as a candidate value. After being corrected by the SOC safety boundary and engine high efficiency zone constraints, it finally obtains a unique optimal α value (such as the example of α=0.6 in the selected content). That is, the multi-objective reward function indirectly drives the model to output the optimal α by defining the "evaluation criteria of optimal α".

[0076] In the above method, the physical constraint verification module executes before the reward function, ensuring the output under any operating condition. All meet the safety and efficiency boundaries, avoiding illegal actions during the reward optimization process. Through online learning, the reward function can dynamically adjust the constraint threshold and the network parameters of the optimization model, thus obtaining a general oil-electricity allocation model after the adjustment is completed.

[0077] If the general oil-electricity distribution model is obtained through the above method, please continue to refer to... Figure 3 This solution extracts shared features across different scenarios (such as the basic decision-making pattern of "low power + high load"), and then generates corresponding scenario-based fuel-electricity allocation models based on these shared features. Specifically, for example, this solution can use the network parameters of a pre-trained general fuel-electricity allocation model as initial values, copying them to the blank network of each corresponding preset scenario. Then, this solution uses a dedicated dataset for each preset scenario (obtainable through filtering training samples) to fine-tune the network parameters for each preset scenario, thereby obtaining the scenario-based fuel-electricity allocation model for each preset scenario.

[0078] In the above implementation method, this scheme uses a proximal policy optimization model combined with physical constraints and a multi-objective reward function to train a general model. The proximal policy optimization model improves the stability and convergence speed of reinforcement learning, while physical constraints and a multi-objective reward function ensure that the model output conforms to the physical characteristics of the vehicle and takes into account multi-dimensional performance objectives.

[0079] In an optional embodiment of this example, after obtaining the target engine power ratio output by the target fuel-electric distribution model, this solution can further employ a model predictive control algorithm to generate a target engine power curve based on the target engine power ratio; control the vehicle to drive according to the target engine power curve; collect the vehicle's engine data and battery data in real time after a preset time period; obtain the engine prediction data and battery prediction data calculated by the multi-objective reward function in the target fuel-electric distribution model; calculate the deviation value based on the engine data, battery data, engine prediction data, and battery prediction data; wherein the deviation value is used to adjust and correct the target fuel-electric distribution model.

[0080] Specifically, this solution can use the Model Predictive Control (MPC) algorithm to smoothly optimize the "real-time output process" of engine power based on the output target engine power ratio, generate a power curve that meets the physical execution requirements, ensure driving smoothness and equipment safety when the decision is implemented, and ultimately achieve "optimal decision α" to "efficient physical execution".

[0081] The vehicle fuel-electricity distribution method described above is based on the deep integration of vehicle-to-everything (V2X) information perception and reinforcement learning dynamic decision-making. It utilizes 5G / C-V2X modules and onboard sensors to collaboratively collect real-time traffic conditions, battery SOC, grid electricity prices, and other data across 25+ dimensions. This data is then used to construct a full-dimensional state vector via edge computing, overcoming the information limitations of traditional systems. The PPO algorithm is employed to process the engine power ratio. The continuous action space is embedded with hard constraints of battery SOC (e.g., 25%-75%) and engine high-efficiency zone (e.g., thermal efficiency >35%). Energy efficiency, economy and safety are balanced through a multi-objective reward function. The hierarchical policy network realizes collaborative decision-making of general model and scenario sub-model. Combining meta-learning and few-shot transfer learning, it can quickly adapt to different scenarios such as urban congestion and mountain climbing.

[0082] like Figure 4 A schematic structural block diagram of a vehicle fuel-electricity distribution device provided in this application is presented. It should be understood that this device is applied to the computing device described above, and this device is used in conjunction with... Figure 1 and 3 The method embodiments executed in the above text correspond to the steps involved in the aforementioned method. The specific functions of the device can be found in the description above, and detailed descriptions are omitted here to avoid repetition. The device includes at least one software functional module that can be stored in memory or embedded in the device's operating system (OS) in the form of software or firmware. Specifically, the device includes: an acquisition module 400, a determination module 410, and an input module 420. The acquisition module 400 is used to acquire the current driving status information of the vehicle, including the vehicle status, driving environment status, and driver status. The determination module 410 is used to determine the target fuel-electric distribution model corresponding to the current driving scenario of the vehicle. The fuel-electric distribution model includes a scenario-based fuel-electric distribution model and a general fuel-electric distribution model. Each scenario-based fuel-electric distribution model is obtained by training on the corresponding driving scenario dataset and the general fuel-electric distribution model. The general fuel-electric distribution model is obtained by reinforcement learning on multiple training samples based on a multi-objective reward function and physical constraints. Each training sample includes the vehicle's historical driving status information, and the physical constraints are used to constrain and limit the range of engine power ratio under different battery states of charge and different engine speeds. The input module 420 is used to input the current driving status information of the vehicle into the target fuel-electric distribution model to obtain the target engine power ratio output by the target fuel-electric distribution model.

[0083] The vehicle fuel-electric distribution device designed above determines the vehicle's driving scenario by integrating multi-dimensional information such as vehicle status, driving environment status, and driver status. It pre-allocates corresponding fuel-electric distribution models based on different driving scenarios, enabling the matching of the target fuel-electric distribution model to the actual determined driving scenario. Then, based on the target fuel-electric distribution model, it achieves fuel-electric distribution under that driving scenario, outputting the target engine power ratio. This achieves scenario-based precise fuel-electric distribution, improving efficient decision-making and rapid adaptation under complex road conditions. Furthermore, the fuel-electric distribution model of this solution incorporates physical constraints to ensure that the battery and engine operate within a safe and efficient range, improving vehicle reliability and economy.

[0084] According to some embodiments of this application, such as Figure 5 As shown, this application provides an electronic device 5, including: a processor 501 and a memory 502. The processor 501 and the memory 502 are interconnected and communicate with each other through a communication bus 503 and / or other forms of connection mechanism (not shown). The memory 502 stores a computer program executable by the processor 501. When the computing device is running, the processor 501 executes the computer program to perform any optional implementation method, such as steps S100 to S130: obtaining the current driving status information of the vehicle; determining the current driving scenario of the vehicle based on the current driving status information of the vehicle; determining the target fuel-electric distribution model corresponding to the current driving scenario of the vehicle based on the current driving scenario of the vehicle; inputting the current driving status information of the vehicle into the target fuel-electric distribution model to obtain the target engine power ratio output by the target fuel-electric distribution model.

[0085] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, performs the method in any of the aforementioned optional implementations.

[0086] The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0087] This application provides a computer program product that, when run on a computer, causes the computer to perform a method in any of the optional implementations.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and not to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. These modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and they should all be covered within the scope of the claims and specification of this application. In particular, as long as there is no structural conflict, the various technical features mentioned in the embodiments can be combined in any way. This application is not limited to the specific embodiments disclosed herein, but includes all technical solutions falling within the scope of the claims.

Claims

1. A vehicle oil-electric distribution method, characterized by, The method comprises: obtaining current driving state information of a vehicle; wherein the driving state information comprises vehicle state, driving environment state and driver state; determining a current driving scenario of the vehicle according to the current driving state information of the vehicle; determining a target oil-electricity distribution model corresponding to the current driving scenario of the vehicle according to the current driving scenario of the vehicle; wherein the oil-electricity distribution model comprises a scenario oil-electricity distribution model and a general oil-electricity distribution model, each of the scenario oil-electricity distribution models is obtained by training corresponding driving scenario data set and the general oil-electricity distribution model, the general oil-electricity distribution model is obtained by reinforcement learning of multiple training samples based on a multi-objective reward function and a physical constraint condition, each of the training samples comprises vehicle historical driving state information, and the physical constraint condition is used to constrain the interval range of engine power ratio under different battery state of charge and different engine speed; inputting the current driving state information of the vehicle into the target oil-electricity distribution model to obtain a target engine power ratio output by the target oil-electricity distribution model.

2. The method of claim 1, wherein, The method further comprises: determining whether the current driving scenario of the vehicle belongs to a target preset scenario according to the current driving scenario of the vehicle; wherein the target preset scenario is any one of multiple preset driving scenarios; if it is determined that the current driving scenario of the vehicle belongs to the target preset scenario, searching for a target scenario oil-electricity distribution model corresponding to the target preset scenario according to the target preset scenario; wherein each scenario oil-electricity distribution model corresponds to a preset driving scenario; determining the target scenario oil-electricity distribution model as the target oil-electricity distribution model corresponding to the current driving scenario of the vehicle.

3. The method of claim 2, wherein, The method further comprises: if it is determined that the current driving scenario of the vehicle does not belong to the target preset scenario, determining the general oil-electricity distribution model as the target oil-electricity distribution model corresponding to the current driving scenario of the vehicle.

4. The method of claim 1, wherein, After obtaining the target engine power ratio output by the target oil-electricity distribution model, the method further comprises: generating a target engine power curve according to the target engine power ratio by using a model predictive control algorithm; controlling the vehicle to drive according to the target engine power curve; after a preset time period, collecting engine data and battery data of the vehicle in real time; obtaining engine prediction data and battery prediction data calculated by a multi-objective reward function in the target oil-electricity distribution model; calculating a deviation value according to the engine data, the battery data, the engine prediction data and the battery prediction data; wherein the deviation value is used to adjust and correct the target oil-electricity distribution model.

5. The method of claim 1, wherein, Before obtaining the current driving state information of the vehicle, the method further comprises: obtaining training samples; wherein the training samples comprise vehicle historical driving state information, and the vehicle historical driving state information comprises vehicle historical state, driving environment historical state and driver historical state; The reinforcement learning is performed on the training sample by using the physical constraint condition and the multi-objective reward function, and the universal oil-electricity distribution model is obtained by using a proximal policy optimization model. Shared features in different scenarios in the universal oil-electricity distribution model are extracted. A driving scene data set in a plurality of pre-classified driving scenes is obtained. According to the shared features in different scenarios in the universal oil-electricity distribution model and the driving scene data set in each driving scene, a scenario oil-electricity distribution model corresponding to each driving scene is trained and obtained.

6. The method of claim 5, wherein, The reinforcement learning is performed on the training sample by using the physical constraint condition and the multi-objective reward function, and the universal oil-electricity distribution model is obtained by using a proximal policy optimization model, and the method comprises the following steps: A high-dimensional state vector of each feature in the training sample is extracted to obtain a full-dimensional state vector corresponding to the training sample. An initial engine power ratio probability distribution of the vehicle in the full-dimensional state is determined according to the full-dimensional state vector corresponding to the training sample by using a proximal policy optimization model. The initial engine power ratio probability distribution is constrained and limited by using the physical constraint condition to obtain a constrained initial engine power ratio probability distribution. A reward and punishment signal corresponding to the constrained initial engine power ratio probability distribution is calculated according to a preset evaluation condition by using a multi-objective reward function. The network parameters of the proximal policy optimization model are optimized and adjusted based on the reward and punishment signal until the optimized and adjusted network parameters converge to obtain the universal oil-electricity distribution model.

7. The method of claim 5, wherein, The physical constraint condition comprises: The engine power ratio of the vehicle is limited by using a first limiting condition based on the current SOC value of the battery. The engine power ratio of the vehicle is limited by using a second limiting condition based on the real-time speed of the engine.

8. The method of claim 7, wherein, The first limiting condition comprises: When the battery SOC is less than a first SOC threshold, the engine power ratio is limited to be greater than a first power ratio threshold. When the battery SOC is greater than a second SOC threshold, the engine power ratio is limited to be less than a second power ratio threshold; wherein the first SOC threshold is less than the second SOC threshold, and the first power ratio threshold is greater than the second power ratio threshold. The second limiting condition is:

9. The method of claim 7, wherein, The engine power ratio of the vehicle is limited to be within a calibrated thermal efficiency interval range corresponding to the real-time speed of the engine; wherein each speed corresponds to a thermal efficiency interval range. The processor executes the computer program to implement the method of any one of claims 1 to 9.

10. An electronic device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor executes the computer program to implement the method of any one of claims 1 to 9.

Images