Engine thermal management control method, device and system, medium and electronic equipment

Abstract

The invention provides an engine thermal management control method, device and system, a medium and electronic equipment, and the method comprises the steps: obtaining driving basic information at the current moment, including first vehicle state information, first engine state information, driving behavior information of a driver and map data information around a vehicle; processing the driving basic information at the current moment based on a driver behavior prediction network model to obtain driving behavior information at each future moment within a preset future duration from the current moment; acquiring second vehicle state information at the current moment as an initial state of the vehicle model, and acquiring second engine state information at the current moment as an initial state of the engine model; and after the driving behavior information of each future moment in the preset future duration is input into the vehicle model and the engine model, the minimum fuel quantity in the preset future duration is used as an optimization target, and the engine thermal management control quantity in the next control period is calculated through a model prediction control algorithm.

Description

Technical Field

[0001] This application relates to the technical field of vehicle engine thermal management, and more specifically, to an engine thermal management control method, device, system, medium, and electronic equipment. Background Technology

[0002] The engine thermal management module plays a crucial role in engine performance. Overheating can disrupt the normal fit and clearance of internal components. If the lubricating oil fails at high temperatures, it can cause parts to seize up. High temperatures also reduce air intake, affecting combustion. Conversely, undercooling the engine can affect oil viscosity, thus impacting lubrication and increasing heat loss.

[0003] Traditional engine thermal management control methods include using temperature sensor input and feedback or feedforward control to regulate the engine temperature close to the optimal operating temperature by controlling the water pump, thermostat, and fan. However, the effectiveness of this method needs improvement, as the actual engine operating temperature fluctuates around the optimal operating temperature. Especially when engine operating conditions change, this can increase control errors and consequently affect fuel economy. Summary of the Invention

[0004] This application provides an engine thermal management control method, device, system, medium, and electronic equipment that can improve control performance with the goal of improving fuel economy.

[0005] The specific technical solution is as follows: In a first aspect, embodiments of this application provide an engine thermal management control method, the method comprising: Obtain basic driving information at the current moment, wherein the basic driving information includes first vehicle status information, first engine status information, driver's driving behavior information, and map data information around the vehicle; The driver behavior prediction network model is used to process the basic driving information at the current moment to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on the basic driving information at multiple historical moments. The current state information of the second vehicle is used as the initial state of the vehicle model, and the current state information of the second engine is used as the initial state of the engine model. After inputting the driving behavior information of each future moment within the preset future time period into the vehicle model and the engine model, the engine thermal management control quantity in the next control cycle is calculated using the model predictive control algorithm with the minimum fuel quantity within the preset future time period as the optimization objective.

[0006] In one possible implementation, the training method for the driver behavior prediction network model includes: Obtain a training set, wherein the training set includes basic driving information at multiple historical moments, and ground truth values ​​of driving behavior information at each moment within the preset future duration starting from each of the historical moments; The initial driver behavior prediction network model is trained using the training set, and the predicted value of driving behavior information is output for each time within the preset future time period starting from each historical time. The current loss value is calculated based on the difference between the true value of the driving behavior information and the predicted value of the driving behavior information at the same moment. The model parameters of the initial driver behavior prediction network model are adjusted using the current loss value, and the driver behavior prediction network model with adjusted parameters is trained until the preset convergence condition is met, thus obtaining the final driver behavior prediction network model.

[0007] In one possible implementation, when the driving behavior information includes multiple driving control behaviors, the true value of the driving behavior information includes multiple true values ​​of driving control behaviors, and the predicted value of the driving behavior information includes multiple predicted values ​​of driving control behaviors. Based on the difference between the true value and the predicted value of the driving behavior information at the same moment, the current loss value is calculated, including: For the true value and predicted value of the driving behavior information at the same moment, calculate the difference between the true value of each driving control behavior and its corresponding predicted value, and calculate the behavior prediction loss value at the same moment based on each difference and the driving control behavior weight corresponding to each difference. The current loss value is calculated based on the time weight of each moment within the preset future time period and the predicted loss value of each moment.

[0008] In one possible implementation, the first vehicle state information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position; and / or, The second vehicle status information includes at least one of vehicle speed, vehicle pose information, heading angle, and yaw rate; and / or, The second engine status information includes at least one of the following: core rotational dynamics status, intake and exhaust system status, torque and combustion status, and thermodynamics and aftertreatment system status.

[0009] In one possible implementation, the engine model includes a thermal management module, an external characteristic model, an engine management system model, and a combustion model, wherein the thermal management module includes a cooling system model and a thermal management actuator model; After inputting the driving behavior information for each future moment within the preset future time period into the vehicle model and the engine model, and using the minimum fuel consumption within the preset future time period as the optimization objective, the engine thermal management control quantity for the next control cycle is calculated using a model predictive control algorithm, including: The brake pedal opening, steering wheel angle, and gear position at each future moment are input into the vehicle model; The vehicle model simulates and calculates the engine speed at the i-th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the i-th future time, and transmits this information to the external characteristic model so that the external characteristic model can transmit the engine speed at the i-th future time to the thermal management actuator model. The thermal management actuator model calculates the internal circulation coolant flow rate, the external circulation coolant flow rate, and the radiator air flow rate at the i-th future time based on the thermal management control quantity calculated by the model predictive control algorithm during the process of minimizing fuel quantity, and transmits these three flow rates to the cooling system model. The engine management system model calculates the fuel injection quantity for the i-th future time based on the coolant temperature at the i-th future time and transmits it to the combustion model. The combustion heat flow of the engine at the (i+1)th future moment is calculated using the combustion model based on the fuel injection quantity and oil temperature at the i-th future moment. The cooling system model calculates the coolant temperature and oil temperature at the (i+1)th future time based on the engine combustion heat flow, the internal circulation coolant flow, the external circulation coolant flow, and the radiator air flow at the (i+1)th future time. These values ​​are then transmitted to the engine management system model and the combustion model, respectively. This process is iteratively repeated until the calculated fuel injection amount for the preset future time is minimized, at which point the corresponding thermal management control value is obtained.

[0010] In one possible implementation, after inputting the driving behavior information for each future moment within the preset future time period into the vehicle model and the engine model, and using the minimum fuel consumption within the preset future time period as the optimization objective, the engine thermal management control quantity for the next control cycle is calculated using a model predictive control algorithm, further comprising: The accelerator pedal opening at each future moment within the preset future time period is input into the external characteristic model; The method for determining the engine torque at the i-th future moment includes: The vehicle model simulates and calculates the engine speed at the (i-1)th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the (i-1)th future time, and transmits it to the external characteristic model. It also simulates and calculates the vehicle speed at the (i-1)th future time and transmits it to the thermal management actuator model. The engine torque at the first future time is determined based on the initial state of the vehicle model. The external characteristic model sends the engine speed at the (i-1)th future time to the thermal management actuator model. After the thermal management actuator model calculates the additional power at the (i-1)th future time based on the thermal management control quantity, engine speed, and vehicle speed calculated by the model predictive control algorithm during the process of minimizing fuel quantity, the external characteristic model calculates the engine torque at the (i-1)th future time based on the accelerator pedal opening, engine speed, and additional power at the (i-1)th future time.

[0011] Secondly, another embodiment of this application provides an engine thermal management control device, the device comprising: The acquisition unit is used to acquire the basic driving information at the current moment, wherein the basic driving information includes first vehicle status information, first engine status information, driver's driving behavior information, and map data information around the vehicle. The prediction unit is used to process the basic driving information at the current moment based on the driver behavior prediction network model to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on the basic driving information at multiple historical moments. An initialization unit is used to obtain the current state information of the second vehicle as the initial state of the vehicle model, and to obtain the current state information of the second engine as the initial state of the engine model. The calculation unit is used to input the driving behavior information of each future moment within the preset future time period into the vehicle model and the engine model, and then use the model predictive control algorithm to calculate the engine thermal management control quantity in the next control cycle with the minimum fuel quantity within the preset future time period as the optimization objective.

[0012] In one possible implementation, the device further includes: Training unit, used to train the driver behavior prediction network model; The training unit includes: The acquisition module is used to acquire a training set, wherein the training set includes basic driving information at multiple historical moments, and truth values ​​of driving behavior information at each moment within the preset future duration starting from each of the historical moments; The training module is used to train the initial driver behavior prediction network model using the training set and output the predicted value of driving behavior information at each time within the preset future time period starting from each historical time. The loss calculation module is used to calculate the current loss value based on the difference between the true value of the driving behavior information and the predicted value of the driving behavior information at the same moment; An adjustment module is used to adjust the model parameters of the initial driver behavior prediction network model using the current loss value; The training module is also used to continue training the driver behavior prediction network model after parameter adjustment until the preset convergence condition is met, so as to obtain the final driver behavior prediction network model.

[0013] In one possible implementation, when the driving behavior information includes multiple driving control behaviors, the true value of the driving behavior information includes multiple true values ​​of driving control behaviors, and the predicted value of the driving behavior information includes multiple predicted values ​​of driving control behaviors. The loss calculation module is used to calculate the difference between the true value of each driving control behavior and its corresponding predicted value for the driving behavior information at the same time, and to calculate the behavior prediction loss value at the same time based on each difference and the driving control behavior weight corresponding to each difference; and to calculate the current loss value based on the time weight of each time within the preset future time and the prediction loss value of each time.

[0014] In one possible implementation, the first vehicle state information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position; and / or, The second vehicle status information includes at least one of vehicle speed, vehicle pose information, heading angle, and yaw rate; and / or, The second engine status information includes at least one of the following: core rotational dynamics status, intake and exhaust system status, torque and combustion status, and thermodynamics and aftertreatment system status.

[0015] In one possible implementation, the engine model includes a thermal management module, an external characteristic model, an engine management system model, and a combustion model, wherein the thermal management module includes a cooling system model and a thermal management actuator model; The computing unit includes: An input module is used to input the brake pedal opening, the steering wheel angle, and the gear position at each future moment into the vehicle model; The first calculation module is used to simulate and calculate the engine speed at the i-th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the i-th future time using the vehicle model, and transmit the result to the external characteristic model so that the external characteristic model can transmit the engine speed at the i-th future time to the thermal management actuator model. The second calculation module is used to calculate the internal circulation coolant flow rate, the external circulation coolant flow rate, and the radiator air flow rate at the i-th future time based on the thermal management control quantity calculated at the i-th future time during the process of minimizing fuel quantity using the thermal management actuator model, and transmit these three flow rates to the cooling system model. The third calculation module is used to calculate the fuel injection quantity at the i-th future time based on the coolant temperature at the i-th future time through the engine management system model and transmit it to the combustion model. The fourth calculation module is used to calculate the engine combustion heat flow at the (i+1)th future time based on the fuel injection quantity and oil temperature at the i-th future time using the combustion model. The fifth calculation module is used to calculate the coolant temperature and oil temperature at the (i+1)th future time based on the engine combustion heat flow, the internal circulation coolant flow, the external circulation coolant flow, and the radiator air flow at the (i+1)th future time, using the cooling system model. These calculations are then transmitted to the engine management system model and the combustion model, respectively. The calculations are iterated continuously until the calculated fuel injection amount for the preset future time is minimized, thus obtaining the corresponding thermal management control quantity.

[0016] In one possible implementation, the input module is further configured to input the accelerator pedal opening at each future moment within the preset future duration into the external characteristic model; The computing unit further includes: A determination module is used to determine the engine torque at the i-th future time moment; The determining module is used for: The vehicle model simulates and calculates the engine speed at the (i-1)th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the (i-1)th future time, and transmits it to the external characteristic model. It also simulates and calculates the vehicle speed at the (i-1)th future time and transmits it to the thermal management actuator model. The engine torque at the first future time is determined based on the initial state of the vehicle model. The external characteristic model sends the engine speed at the (i-1)th future time to the thermal management actuator model. After the thermal management actuator model calculates the additional power at the (i-1)th future time based on the thermal management control quantity, engine speed, and vehicle speed calculated by the model predictive control algorithm during the process of minimizing fuel quantity, the external characteristic model calculates the engine torque at the (i-1)th future time based on the accelerator pedal opening, engine speed, and additional power at the (i-1)th future time.

[0017] Thirdly, another embodiment of this application provides an engine thermal management control system, the system including a driver behavior prediction network model, a vehicle model and an engine model; The driver behavior prediction network model is used to acquire basic driving information at the current moment and process the basic driving information at the current moment to obtain driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on driving basic information at multiple historical moments. The basic driving information includes first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle. The vehicle model is used to obtain the current state information of the second vehicle as the initial state; The engine model is used to obtain the current state information of the second engine as the initial state, and combined with the initialized vehicle model, with the minimum fuel quantity within the preset future time period as the optimization objective, and using the model predictive control algorithm and the driving behavior information of each future time within the preset future time period, to calculate the engine thermal management control quantity in the next control cycle.

[0018] In one possible implementation, the first vehicle state information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position; and / or, The second vehicle status information includes at least one of vehicle speed, vehicle pose information, heading angle, and yaw rate; and / or, The second engine status information includes at least one of the following: core rotational dynamics status, intake and exhaust system status, torque and combustion status, and thermodynamics and aftertreatment system status.

[0019] In one possible implementation, the engine model includes a thermal management module, an external characteristic model, an engine management system model, and a combustion model, wherein the thermal management module includes a cooling system model and a thermal management actuator model; The vehicle model is used to obtain the brake pedal opening, steering wheel angle and gear position at each future time, and to simulate and calculate the engine speed at the i-th future time based on the engine torque, brake pedal opening, steering wheel angle and gear position at the i-th future time, and transmit it to the external characteristic model. The external characteristic model is used to transmit the engine speed at the i-th future moment to the thermal management actuator model; The thermal management actuator model is used to calculate the internal circulation coolant flow rate, the external circulation coolant flow rate, and the radiator air flow rate at the i-th future time based on the thermal management control quantity calculated at the i-th future time during the process of minimizing fuel quantity using the model predictive control algorithm, and the engine speed at the i-th future time, and transmit these three flow rates to the cooling system model. The engine management system model is used to calculate the fuel injection quantity at the i-th future time based on the coolant temperature at the i-th future time and transmit it to the combustion model. The combustion model is used to calculate the engine combustion heat flow at the (i+1)th future time based on the fuel injection quantity and the oil temperature at the i-th future time. The cooling system model is used to calculate the coolant temperature and oil temperature at the (i+1)th future time based on the engine combustion heat flow, the internal circulation coolant flow, the external circulation coolant flow, and the radiator air flow at the (i+1)th future time, and transmit them to the engine management system model and the combustion model, respectively. The calculation is iterated in this way until the fuel injection quantity for the preset future time is minimized, and the corresponding thermal management control quantity is obtained.

[0020] In one possible implementation, the external characteristic model is used to obtain the accelerator pedal opening at each future moment within the preset future time period; The vehicle model is used to simulate and calculate the engine speed at the (i-1)th future time and transmit it to the external characteristic model, and to simulate and calculate the vehicle speed at the (i-1)th future time and transmit it to the thermal management actuator model, based on the engine torque at the (i-1)th future time, the brake pedal opening at the (i-1)th future time, the steering wheel angle at the (i-1)th future time, and the gear position at the (i-1)th future time. The engine torque at the first future time is determined based on the initial state of the vehicle model. The external characteristic model is used to send the engine speed at the (i-1)th future time step to the thermal management actuator model; The thermal management actuator model is used to calculate the additional power at the (i-1)th future time based on the thermal management control quantity, engine speed, and vehicle speed calculated by the model predictive control algorithm during the process of minimizing fuel quantity. The external characteristic model is also used to calculate the engine torque at the i-th future time based on the accelerator pedal opening at the (i-1)-th future time, the engine speed at the (i-1)-th future time, and the additional power at the (i-1)-th future time.

[0021] Fourthly, another embodiment of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method as described in any embodiment of the first aspect.

[0022] Fifthly, another embodiment of this application provides an electronic device, the electronic device comprising: One or more processors; Storage device for storing one or more programs. Wherein, when the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method as described in any embodiment of the first aspect.

[0023] In a sixth aspect, embodiments of this application provide a computer program product containing instructions that, when executed on a computer or processor, cause the computer or processor to perform the method described in any possible implementation of the first aspect.

[0024] As can be seen from the above, the engine thermal management control method, device, system, medium, and electronic equipment provided in this application embodiment can first acquire the basic driving information at the current moment (including first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle), then process the basic driving information at the current moment based on a pre-trained driver behavior prediction network model to obtain driving behavior information for each future moment within a preset future time period starting from the current moment, then acquire the second vehicle state information at the current moment as the initial state of the vehicle model, and acquire the second engine state information at the current moment as the initial state of the engine model, and finally input the driving behavior information for each future moment within the preset future time period into the vehicle model and the engine model, and use the model predictive control algorithm to calculate the engine thermal management control quantity in the next control cycle with the minimum fuel quantity within the preset future time period as the optimization objective. Therefore, the embodiments of this application not only aim to reduce fuel consumption and improve fuel economy, but also automatically predict the engine thermal management control quantity in the next control cycle to better improve the impact of engine thermal management control on fuel economy. Furthermore, when predicting future driving behavior information, the influence of map data information around the vehicle on engine operating conditions, such as vehicle climbing conditions, turning conditions, intersection deceleration conditions, and speed limit conditions, can be considered. Thus, regardless of how the operating conditions change, the fuel consumption can be minimized, further improving the impact of engine thermal management control on fuel economy.

[0025] Of course, implementing any product or method of this application does not necessarily require achieving all of the advantages described above at the same time. Attached Figure Description

[0026] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0027] Figure 1 A flowchart illustrating a driver behavior prediction network model training method provided in an embodiment of this application; Figure 2 A flowchart illustrating a driving behavior prediction method provided in an embodiment of this application; Figure 3 A schematic flowchart of an engine thermal management control method provided in an embodiment of this application; Figure 4 A schematic flowchart illustrating another engine thermal management control method provided in an embodiment of this application; Figure 5 A flowchart illustrating another engine thermal management control method provided in this application embodiment; Figure 6 A block diagram of an engine thermal management control device provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of an engine thermal management control system provided in an embodiment of this application. Detailed Implementation

[0028] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0029] It should be noted that the terms "comprising" and "having," and any variations thereof, in the embodiments and accompanying drawings of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0030] Figure 1 This application provides a flowchart illustrating a driver behavior prediction network model training method. This method can be applied to electronic devices, such as computers, and mainly includes: S110: Obtain the training set, which includes basic driving information at multiple historical moments, and ground truth values ​​of driving behavior information at each moment within a preset future duration starting from each historical moment.

[0031] The basic driving information includes the first vehicle status information, the first engine status information, the driver's driving behavior information, and the map data information around the vehicle. The driving behavior information includes the driver's control behavior of the vehicle during the driving process.

[0032] The first vehicle status information includes at least one of the following: vehicle speed, acceleration, sideslip angle, yaw rate, etc.; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, Map data information includes at least one of the following: road elevation, road curvature, traffic lights, traffic signs, and the presence or absence of intersections within a preset distance (e.g., 100m) ahead of the vehicle; and / or, Driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, gear position, etc.

[0033] In this embodiment, the first vehicle status information, the first engine status information, and the driving behavior information can all be detected using corresponding sensors (e.g., acceleration can be detected by an acceleration sensor) and obtained from the vehicle's CAN (Controller Area Network) bus. The engine thermal management control method provided in this embodiment can be deployed in an ECU controller with a CAN interface, or in other controllers with a CAN interface. When deployed in an ECU (Electronic Control Unit) controller, the ECU controller can obtain the first vehicle status information, the first engine status information, and the driving behavior information via the CAN bus.

[0034] This application embodiment can pre-store high-precision maps in controllers such as ECUs, and then query map data information around the vehicle from the high-precision map using the vehicle's current location information. If no high-precision map is pre-stored in the vehicle, the vehicle can first upload its current location information to the cloud via a mobile communication module, and then the cloud will publish map data information near that location. The vehicle's current location information can be obtained from the positioning results of positioning systems such as GPS (Global Positioning System), BDS (Beidou Navigation Satellite System), GNSS (Global Navigation Satellite System), or a combined inertial navigation system. The combined inertial navigation system includes a GPS positioning system / BDS system / GNSS system and an INS (Inertial Navigation System) orientation and navigation system. The high-precision map can be constructed by fusing data collected by high-precision positioning equipment, LiDAR, cameras, and other devices.

[0035] It should be noted that when constructing the training set, basic driving information within a historical time period can be collected. Besides the last preset future duration within this historical time period, for each historical moment, the driving behavior information within the basic driving information for the next preset future duration can be compiled as the ground truth value of driving behavior information for each moment within the preset future duration starting from that historical moment. For example, when a historical time period is 24 seconds, each moment is 1 second, and the preset future duration is 7 seconds, the ground truth value of driving behavior information for the next 7 seconds corresponding to the 1st second includes: driving behavior information from the basic driving information of the 2nd to the 8th second; the ground truth value of driving behavior information for the next 7 seconds corresponding to the 2nd second includes: driving behavior information from the basic driving information of the 3rd to the 9th second; the ground truth value of driving behavior information for the next 7 seconds corresponding to the 3rd second includes: driving behavior information from the basic driving information of the 4th to the 10th second, and so on, until the ground truth value of driving behavior information for the next 7 seconds corresponding to the 17th second includes: driving behavior information from the basic driving information of the 18th to the 24th second.

[0036] Furthermore, the driving behavior information in the training set does not need to be completely continuous in time, as long as each historical moment is continuous with its corresponding future preset duration. In other words, all driving behavior information in the training set can consist of multiple time periods, and these time periods do not need to be continuous or overlap. For example, when each moment is 1 second and the preset future duration is 7 seconds, if we obtain the basic driving information for the historical time periods 1s-8s, 11s-18s, 20s-27s, and 32s-39s, the training set only needs to include: the basic driving information for the 1st second, the driving behavior information (true value) for the 2nd to 8th seconds, the basic driving information for the 11th second, the driving behavior information (true value) for the 12th to 18th seconds, the basic driving information for the 20th second, the driving behavior information (true value) for the 21st to 27th seconds, the basic driving information for the 32nd second, and the driving behavior information (true value) for the 33rd to 39th seconds.

[0037] S120: Train the initial driver behavior prediction network model using the training set, and output the predicted value of driving behavior information at each time within a preset future time period starting from each historical time.

[0038] This application does not limit the network structure of the driver behavior prediction network model, as long as it can be trained to produce driving behavior information that meets the accuracy requirements. The neural network model can be a fully connected neural network, including an input layer, several hidden layers, and an output layer. For example, during normal driving, if the elevation of the road ahead gradually increases, it indicates that an uphill climb is imminent, and the driver's subsequent control is likely to increase the accelerator pedal opening. Similarly, during normal driving, if the curvature of the road ahead increases, it indicates that a curve is imminent, and the driver's subsequent control is likely to increase the steering wheel angle. These relationships may be non-linear.

[0039] After obtaining the initial driver behavior prediction network model, the training set can be input into the initial driver behavior prediction network model for training. Through several hidden layers, the output layer outputs the predicted values ​​of driving behavior information for each time within a preset future time period starting from each historical time.

[0040] S130: Calculate the current loss value based on the difference between the true value of driving behavior information and the predicted value of driving behavior information at the same time.

[0041] When driving behavior information includes multiple driving control behaviors, the true value of driving behavior information includes multiple true values ​​of driving control behaviors, and the predicted value of driving behavior information includes multiple predicted values ​​of driving control behaviors.

[0042] In this case, the specific calculation process for the current loss value may include: For the true value and predicted value of driving behavior information at the same moment, calculate the difference between the true value of each driving control behavior and its corresponding predicted value, and calculate the behavior prediction loss value at the same moment based on each difference and the driving control behavior weight corresponding to each difference; calculate the current loss value based on the time weight of each moment within a preset future time and the predicted loss value of each moment.

[0043] The weights of different driving control behaviors can be the same or different. When the weights are the same, the weight can be the reciprocal of the number of driving control behaviors at a given moment. When the weights are different, different weights can be assigned to different driving control behaviors based on their importance, so that the higher the importance, the greater the weight.

[0044] The weights of different moments can be the same or different. When the weights are the same, the weight can be the reciprocal of the number of moments included in the preset future duration. When the weights are different, different weights can be assigned to different moments based on their importance; for example, the earlier a moment is, the greater its weight.

[0045] The method for calculating the behavior prediction loss value at the same time moment, based on each difference and the corresponding driving control behavior weight, can be as follows: For multiple driving control behaviors at the same time moment, the absolute value of each difference and its corresponding driving control behavior weight are directly used for weighted calculation to obtain the behavior prediction loss value at the same time moment. For example, the behavior prediction loss value at time i = driving control behavior weight 1 | accelerator pedal opening true value i - accelerator pedal opening predicted value i| + driving control behavior weight 2 | brake pedal opening true value i - brake pedal opening predicted value i| + driving control behavior weight 3 | steering wheel angle true value i - steering wheel angle predicted value i| + driving control behavior weight 4 | gear position true value i - accelerator gear position predicted value i|.

[0046] Based on each difference and its corresponding driving control behavior weight, another method for calculating the behavior prediction loss value at the same time step is as follows: For multiple driving control behaviors at the same time step, first calculate the square of each difference, then use the square of each difference and its corresponding driving control behavior weight to perform a weighted calculation to obtain the behavior prediction loss value at the same time step. For example, the behavior prediction loss value at time step i = driving control behavior weight 1 (accelerator pedal opening true value i - accelerator pedal opening predicted value i) 2 +Driver control behavior weight 2 (true value of brake pedal opening i - predicted value of brake pedal opening i) 2 +Driving control behavior weight 3 (True value of steering wheel angle i - Predicted value of steering wheel angle i) 2 +Driving control behavior weight 4 (Gear true value i - Throttle gear prediction value i) 2 .

[0047] The method for calculating the current loss value, based on the time weight of each moment within a preset future time period and the predicted loss value of each moment, can include: performing a weighted calculation based on the time weight of each moment within the preset future time period and the predicted loss value of each moment to obtain the current loss value. For example, if the preset future time period includes n moments, then the current loss value = time weight 1 * predicted loss value 1 + time weight 2 * predicted loss value 2 + ... + time weight n * predicted loss value n.

[0048] S140: Adjust the model parameters of the initial driver behavior prediction network model using the current loss value, and continue to train the driver behavior prediction network model with adjusted parameters until the preset convergence condition is met, thus obtaining the final driver behavior prediction network model.

[0049] The preset convergence conditions include that the current loss value is less than or equal to a loss threshold, or that the number of training epochs has reached a threshold. If the current state does not meet the preset convergence conditions, backpropagation can be used to calculate the gradient of the current loss value, and then the model parameters of the initial driver behavior prediction network model can be updated based on the calculated gradient. This process of forward propagation (i.e., model prediction), calculating the current loss value, backpropagation, and parameter updating is repeated multiple times, each time using different data samples, until the current loss value of the model on the training set is less than or equal to the loss threshold, or the number of training epochs has reached a threshold. At this point, model training stops, and the resulting model is the final driver behavior prediction network model.

[0050] The driver behavior prediction network model training method provided in this application embodiment can first construct a training set including basic driving information at multiple historical moments, and the true values ​​of driving behavior information at each moment within the preset future time period starting from each historical moment. The basic driving information includes first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle. Then, the driver behavior prediction network model is trained using the training set. Since the map data information around the vehicle is integrated when training the driver behavior prediction network model, the accuracy of the driver behavior prediction network model in predicting driver behavior is improved.

[0051] Based on the above embodiments, another embodiment of this application provides a driving behavior prediction method. This method can be applied to electronic devices, specifically to vehicles, or to servers, such as... Figure 2 As shown, the method includes: S210: Obtain the basic driving information at the current moment, including the first vehicle status information, the first engine status information, the driver's driving behavior information, and the map data information around the vehicle.

[0052] The first vehicle status information includes at least one of the following: vehicle speed, acceleration, sideslip angle, yaw rate, etc.; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, Map data information includes at least one of the following: road elevation, road curvature, traffic lights, traffic signs, and the presence or absence of intersections within a preset distance (e.g., 100m) ahead of the vehicle; and / or, Driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, gear position, etc.

[0053] The driving basic information obtained in this step at the current moment is conceptually the same as the driving basic information at historical moments in the training set obtained in step S110, except that the data collection time is different. Therefore, for other details about driving basic information, please refer to the detailed explanation of step S110, which will not be repeated here.

[0054] It should be noted that each future moment within the preset future duration starting from the current moment does not include the current moment, but only future moments after the current moment.

[0055] S220: Based on the driver behavior prediction network model, the basic driving information at the current moment is processed to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment.

[0056] The driver behavior prediction network model is a neural network model trained based on driving information from multiple historical time points. For details on the training method of the driver behavior prediction network model, please refer to [link to relevant documentation]. Figure 1 The embodiments shown will not be described in detail here.

[0057] When the vehicle has a pre-trained driver behavior prediction network model, after obtaining the basic driving information at the current moment, the vehicle can directly input the basic driving information at the current moment into the driver behavior prediction network model, and output the driving behavior information at each future moment within a preset future time period starting from the current moment. When the vehicle does not have a pre-trained driver behavior prediction network model, the vehicle can report the basic driving information at the current moment to the server, and the server will input the basic driving information at the current moment into the driver behavior prediction network model, output the driving behavior information at each future moment within a preset future time period starting from the current moment, and feed it back to the vehicle.

[0058] After obtaining driving behavior information for each future moment within a preset future timeframe, the vehicle can perform subsequent engine thermal management. Alternatively, it can use the predicted future driving behavior information to determine if a collision risk exists; if so, an immediate warning is issued. To allow the driver reaction time, the preset future timeframe in this case is longer than that used in engine thermal management. Furthermore, the method for predicting driving behavior information in this application can also be applied to other scenarios to improve the driving experience in those situations.

[0059] The driving behavior prediction method provided in this application first obtains the basic driving information at the current moment, and then processes the basic driving information at the current moment based on a driver behavior prediction network model trained in advance based on the basic driving information at multiple historical moments. This predicts the driving behavior information at each future moment within a preset future time period starting from the current moment. Since the prediction of future driving behavior information takes into account the influence of map data information around the vehicle on engine operating conditions, such as vehicle climbing conditions, turning conditions, intersection deceleration conditions, and speed limit conditions, the fuel quantity can be minimized regardless of how the operating conditions change, further improving the impact of engine thermal management control on fuel economy.

[0060] Based on the above embodiments, another embodiment of this application provides an engine thermal management control method, which can be applied to electronic devices, specifically vehicles, such as... Figure 3 and Figure 4 As shown, the method mainly includes: S310: Obtain the basic driving information at the current moment, including the first vehicle status information, the first engine status information, the driver's driving behavior information, and the map data information around the vehicle.

[0061] The driving basic information obtained in this step at the current moment is conceptually the same as the driving basic information at historical moments in the training set obtained in step S110, except that the data collection time is different. Therefore, for other details about driving basic information, please refer to the detailed explanation of step S110, which will not be repeated here.

[0062] It should be noted that each future moment within the preset future duration starting from the current moment does not include the current moment, but only future moments after the current moment.

[0063] S320: Based on the driver behavior prediction network model, the basic driving information at the current moment is processed to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment.

[0064] The driver behavior prediction network model is a neural network model trained based on driving information from multiple historical time points. For details on the training method of the driver behavior prediction network model, please refer to [link to relevant documentation]. Figure 1 The embodiments shown will not be described in detail here.

[0065] S330: Obtain the current state information of the second vehicle as the initial state of the vehicle model, and obtain the current state information of the second engine as the initial state of the engine model.

[0066] The vehicle model includes a vehicle kinematics model and / or a vehicle dynamics model, while the engine model is a mathematical model that simulates the operation of a vehicle engine, rather than a neural network model.

[0067] Vehicle models on the market may differ, and the specific vehicle state information used by a vehicle model depends on the model itself. Furthermore, the vehicle state information used by a vehicle model may not be the same as that used by a driver behavior prediction network model. Therefore, to distinguish between the vehicle state information used by the driver behavior prediction network model and the vehicle state information used by the vehicle model, this application uses "first vehicle state information" and "second vehicle state information" for description. "First" and "second" do not indicate order, but only a difference in name. For example, the second vehicle state information includes at least one of vehicle speed, vehicle pose information, heading angle, and yaw rate. When the vehicle model is complex, the second vehicle state information may also include suspension formation, vehicle pitch angle, vehicle roll angle, and wheel state (such as wheel rotation speed, wheel slip ratio / slip angle), etc. The second engine state information includes at least one of core rotational dynamics state, intake and exhaust system state, torque and combustion state, and thermodynamics and aftertreatment system state. Among them, the core rotational dynamics state includes engine speed, crankshaft angle, etc.; the intake and exhaust system state includes intake manifold pressure, intake manifold temperature, atmospheric pressure, atmospheric temperature, etc.; torque and combustion state includes indicated torque / braking torque, air-fuel ratio (i.e., the mass ratio of air to fuel in the cylinder), ignition advance angle / injection timing, etc.; and thermodynamic and after-treatment system state includes engine coolant temperature, engine oil temperature, cylinder wall temperature, three-way catalytic converter temperature, etc.

[0068] Similarly, in order to distinguish between the engine state information used by the driver behavior prediction network model and the engine state information used by the engine model, this application embodiment uses "first engine state information" and "second engine state information" for description.

[0069] S340: After inputting the driving behavior information of each future moment within the preset future time period into the vehicle model and engine model, the engine thermal management control quantity in the next control cycle is calculated using the model predictive control algorithm with the minimum fuel quantity within the preset future time period as the optimization objective.

[0070] Among them, engine thermal management control parameters include the on / off state of the water pump, the opening degree of the thermostat, and the speed of the fan.

[0071] Model Predictive Control (MPC) is a special type of control. Its current control action is obtained at each sample instant by solving a finite-time open-loop optimal control problem. The current state of the process serves as the initial state of the optimal control problem, and the resulting optimal control sequence only implements the first control action. This is the biggest difference between MPC and algorithms that use pre-computed control laws. Essentially, MPC solves an open-loop optimal control problem. Its underlying concept is independent of the specific model, but its implementation is model-dependent.

[0072] like Figure 5 As shown, the engine model includes a thermal management module, an external characteristic model, an EMS (Engine Management System) model, and a combustion model. The thermal management module includes a cooling system model and a thermal management actuator model.

[0073] The driver behavior prediction network model predicts the accelerator pedal opening at future moments, which can be input into the external characteristic model of the engine model. The predicted brake pedal opening, steering wheel angle, and gear position at future moments can be input into the vehicle model. The vehicle model, based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the i-th future moment, simulates and calculates the engine speed at the i-th future moment, which is then transmitted to the external characteristic model. The external characteristic model then transmits the engine speed at the i-th future moment to the thermal management actuator model. The thermal management actuator model, based on the thermal management control quantity calculated using the model predictive control algorithm during fuel minimization, and the engine speed at the i-th future moment, calculates the internal coolant flow rate, external coolant flow rate, and... The radiator airflow at time i in the future is calculated, and these three airflow parameters are transmitted to the cooling system model. The EMS model calculates the fuel injection quantity at time i in the future based on the coolant temperature at time i in the future and transmits it to the combustion model. The combustion model calculates the engine combustion heat flow at time i+1 in the future based on the fuel injection quantity and oil temperature at time i in the future. The cooling system model calculates the coolant temperature and oil temperature at time i+1 in the future based on the engine combustion heat flow, the internal circulation coolant flow, the external circulation coolant flow, and the radiator airflow at time i in the future, and transmits them to the EMS model and the combustion model, respectively. This process is repeated iteratively until the calculated fuel injection quantity for the preset future time is minimized, at which point the corresponding thermal management control quantity is obtained.

[0074] The method for determining the engine torque at the i-th future time step includes: simulating and calculating the engine speed at the i-1-th future time step using the vehicle model based on the engine torque at the i-1-th future time step, the brake pedal opening at the i-1-th future time step, the steering wheel angle at the i-1-th future time step, and the gear position at the i-1-th future time step, and transmitting this calculation to the external characteristic model; simulating and calculating the vehicle speed at the i-1-th future time step and transmitting this calculation to the thermal management actuator model; sending the engine speed at the i-1-th future time step to the thermal management actuator model via the external characteristic model; calculating the additional power at the i-1-th future time step based on the thermal management control quantity, engine speed, and vehicle speed calculated by the thermal management actuator model during the process of minimizing fuel quantity using the model predictive control algorithm; and finally, calculating the engine torque at the i-1-th future time step based on the accelerator pedal opening, engine speed, and additional power at the i-1-th future time step.

[0075] The engine torque at the first future time step is determined based on the initial state of the vehicle model. The coolant temperature and oil temperature at the first future time step are also determined based on the initial state of the engine model.

[0076] The engine thermal management control method provided in this application first acquires the basic driving information at the current moment (including first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle), then processes the basic driving information at the current moment based on a pre-trained driver behavior prediction network model to obtain driving behavior information for each future moment within a preset future time period starting from the current moment. Then, it acquires the second vehicle state information at the current moment as the initial state of the vehicle model and the second engine state information at the current moment as the initial state of the engine model. Finally, it inputs the driving behavior information for each future moment within the preset future time period into the vehicle model and the engine model, and uses the minimum fuel quantity within the preset future time period as the optimization objective to calculate the engine thermal management control quantity in the next control cycle using a model predictive control algorithm. Therefore, the embodiments of this application not only aim to reduce fuel consumption and improve fuel economy, but also automatically predict the engine thermal management control quantity in the next control cycle to better improve the impact of engine thermal management control on fuel economy. Furthermore, when predicting future driving behavior information, the influence of map data information around the vehicle on engine operating conditions, such as vehicle climbing conditions, turning conditions, intersection deceleration conditions, and speed limit conditions, can be considered. Thus, regardless of how the operating conditions change, the fuel consumption can be minimized, further improving the impact of engine thermal management control on fuel economy.

[0077] Based on the above embodiments, another embodiment of this application provides a driver behavior prediction network model training device, which can be applied to electronic devices, specifically computers, and the device includes: An acquisition unit is used to acquire a training set, wherein the training set includes basic driving information at multiple historical moments, and true values ​​of driving behavior information at each moment within a preset future time period starting from each of the historical moments. The basic driving information includes first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle. The training unit is used to train the initial driver behavior prediction network model using the training set and output the predicted value of driving behavior information at each time within the preset future time period starting from each historical time. The loss calculation unit is used to calculate the current loss value based on the difference between the true value of the driving behavior information and the predicted value of the driving behavior information at the same time. An adjustment unit is used to adjust the model parameters of the initial driver behavior prediction network model using the current loss value; The training unit is also used to continue training the driver behavior prediction network model after parameter adjustment until the preset convergence condition is met, so as to obtain the final driver behavior prediction network model.

[0078] In one possible implementation, when the driving behavior information includes multiple driving control behaviors, the true value of the driving behavior information includes multiple true values ​​of driving control behaviors, and the predicted value of the driving behavior information includes multiple predicted values ​​of driving control behaviors. The loss calculation unit includes: The difference calculation module is used to calculate the difference between the true value of each driving control behavior and its corresponding predicted value for the driving behavior information at the same time. The weighted calculation module is used to calculate the behavior prediction loss value at the same moment based on each difference and the driving control behavior weight corresponding to each difference, and to calculate the current loss value based on the time weight of each moment within the preset future time and the prediction loss value of each moment.

[0079] In one possible implementation, the first vehicle state information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position.

[0080] The driver behavior prediction network model training device provided in this application embodiment can first construct a training set including basic driving information at multiple historical moments, and the true values ​​of driving behavior information at each moment within the preset future time period starting from each historical moment. The basic driving information includes first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle. Then, the driver behavior prediction network model is trained using the training set. Since the map data information around the vehicle is integrated when training the driver behavior prediction network model, the accuracy of the driver behavior prediction network model in predicting driver behavior is improved.

[0081] Based on the above embodiments, another embodiment of this application provides a driving behavior prediction device, which can be applied to electronic devices, specifically to vehicles, and the device includes: The acquisition unit is used to acquire the basic driving information at the current moment, wherein the basic driving information includes first vehicle status information, first engine status information, driver's driving behavior information, and map data information around the vehicle. The prediction unit is used to process the basic driving information at the current moment based on the driver behavior prediction network model to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on the basic driving information at multiple historical moments.

[0082] In one possible implementation, the device further includes: Training unit, used to train the driver behavior prediction network model; The training unit includes: The acquisition module is used to acquire a training set, wherein the training set includes basic driving information at multiple historical moments, and truth values ​​of driving behavior information at each moment within the preset future duration starting from each of the historical moments; The training module is used to train the initial driver behavior prediction network model using the training set and output the predicted value of driving behavior information at each time within the preset future time period starting from each historical time. The loss calculation module is used to calculate the current loss value based on the difference between the true value of the driving behavior information and the predicted value of the driving behavior information at the same moment; An adjustment module is used to adjust the model parameters of the initial driver behavior prediction network model using the current loss value; The training module is also used to continue training the driver behavior prediction network model after parameter adjustment until the preset convergence condition is met, so as to obtain the final driver behavior prediction network model.

[0083] In one possible implementation, when the driving behavior information includes multiple driving control behaviors, the true value of the driving behavior information includes multiple true values ​​of driving control behaviors, and the predicted value of the driving behavior information includes multiple predicted values ​​of driving control behaviors. The loss calculation module is used to calculate the difference between the true value of each driving control behavior and its corresponding predicted value for the driving behavior information at the same time, and to calculate the behavior prediction loss value at the same time based on each difference and the driving control behavior weight corresponding to each difference; and to calculate the current loss value based on the time weight of each time within the preset future time and the prediction loss value of each time.

[0084] In one possible implementation, the first vehicle state information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position.

[0085] The driving behavior prediction device provided in this application embodiment can first acquire the basic driving information at the current moment, and then process the basic driving information at the current moment based on the driver behavior prediction network model trained in advance based on the basic driving information at multiple historical moments, and predict the driving behavior information at each future moment within a preset future time period starting from the current moment. Since the prediction of future driving behavior information takes into account the influence of map data information around the vehicle on engine operating conditions, such as vehicle climbing conditions, turning conditions, intersection deceleration conditions, speed limit conditions, etc., the fuel quantity can be minimized no matter how the operating conditions change, further improving the impact of engine thermal management control on fuel economy.

[0086] Based on the above embodiments, another embodiment of this application provides an engine thermal management control device, which can be applied to electronic devices, specifically to vehicles, such as... Figure 6 As shown, the device includes: The acquisition unit 410 is used to acquire the driving basic information at the current moment, wherein the driving basic information includes first vehicle status information, first engine status information, driver's driving behavior information and map data information around the vehicle. The prediction unit 420 is used to process the basic driving information at the current moment based on the driver behavior prediction network model to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on the basic driving information at multiple historical moments. The initialization unit 430 is used to obtain the current state information of the second vehicle as the initial state of the vehicle model, and to obtain the current state information of the second engine as the initial state of the engine model. The calculation unit 440 is used to input the driving behavior information of each future moment within the preset future time period into the vehicle model and the engine model, and then use the model predictive control algorithm to calculate the engine thermal management control quantity in the next control cycle with the minimum fuel quantity within the preset future time period as the optimization objective.

[0087] In one possible implementation, the device further includes: Training unit, used to train the driver behavior prediction network model; The training unit includes: The acquisition module is used to acquire a training set, wherein the training set includes basic driving information at multiple historical moments, and truth values ​​of driving behavior information at each moment within the preset future duration starting from each of the historical moments; The training module is used to train the initial driver behavior prediction network model using the training set and output the predicted value of driving behavior information at each time within the preset future time period starting from each historical time. The loss calculation module is used to calculate the current loss value based on the difference between the true value of the driving behavior information and the predicted value of the driving behavior information at the same moment; An adjustment module is used to adjust the model parameters of the initial driver behavior prediction network model using the current loss value; The training module is also used to continue training the driver behavior prediction network model after parameter adjustment until the preset convergence condition is met, so as to obtain the final driver behavior prediction network model.

[0088] In one possible implementation, when the driving behavior information includes multiple driving control behaviors, the true value of the driving behavior information includes multiple true values ​​of driving control behaviors, and the predicted value of the driving behavior information includes multiple predicted values ​​of driving control behaviors. The loss calculation module is used to calculate the difference between the true value of each driving control behavior and its corresponding predicted value for the driving behavior information at the same time, and to calculate the behavior prediction loss value at the same time based on each difference and the driving control behavior weight corresponding to each difference; and to calculate the current loss value based on the time weight of each time within the preset future time and the prediction loss value of each time.

[0089] In one possible implementation, the plurality of historical moments are consecutive historical moments.

[0090] In one possible implementation, the first vehicle state information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position; and / or, The second vehicle status information includes at least one of vehicle speed, vehicle pose information, heading angle, and yaw rate; and / or, The second engine status information includes at least one of the following: core rotational dynamics status, intake and exhaust system status, torque and combustion status, and thermodynamics and aftertreatment system status.

[0091] In one possible implementation, the engine model includes a thermal management module, an external characteristic model, an engine management system model, and a combustion model, wherein the thermal management module includes a cooling system model and a thermal management actuator model; The computing unit includes: The input module is used to input the accelerator pedal opening at each future moment within the preset future time period into the external characteristic model, and to input the brake pedal opening, steering wheel angle and gear position at each future moment into the vehicle model. The first calculation module is used to simulate and calculate the engine speed at the i-th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the i-th future time using the vehicle model, and transmit the result to the external characteristic model so that the external characteristic model can transmit the engine speed at the i-th future time to the thermal management actuator model. The second calculation module is used to calculate the internal circulation coolant flow rate, the external circulation coolant flow rate, and the radiator air flow rate at the i-th future time based on the thermal management control quantity calculated at the i-th future time during the process of minimizing fuel quantity using the thermal management actuator model, and transmit these three flow rates to the cooling system model. The third calculation module is used to calculate the fuel injection quantity at the i-th future time based on the coolant temperature at the i-th future time through the engine management system model and transmit it to the combustion model. The fourth calculation module is used to calculate the engine combustion heat flow at the (i+1)th future time based on the fuel injection quantity and oil temperature at the i-th future time using the combustion model. The fifth calculation module is used to calculate the coolant temperature and oil temperature at the (i+1)th future time based on the engine combustion heat flow, the internal circulation coolant flow, the external circulation coolant flow, and the radiator air flow at the (i+1)th future time, using the cooling system model. These calculations are then transmitted to the engine management system model and the combustion model, respectively. The calculations are iterated continuously until the calculated fuel injection amount for the preset future time is minimized, thus obtaining the corresponding thermal management control quantity.

[0092] In one possible implementation, the computing unit further includes: A determination module is used to determine the engine torque at the i-th future time moment; The determining module is used for: The vehicle model simulates and calculates the engine speed at the (i-1)th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the (i-1)th future time, and transmits it to the external characteristic model. It also simulates and calculates the vehicle speed at the (i-1)th future time and transmits it to the thermal management actuator model. The engine torque at the first future time is determined based on the initial state of the vehicle model. The external characteristic model sends the engine speed at the (i-1)th future time to the thermal management actuator model. After the thermal management actuator model calculates the additional power at the (i-1)th future time based on the thermal management control quantity, engine speed, and vehicle speed calculated by the model predictive control algorithm during the process of minimizing fuel quantity, the external characteristic model calculates the engine torque at the (i-1)th future time based on the accelerator pedal opening, engine speed, and additional power at the (i-1)th future time.

[0093] The engine thermal management control device provided in this application embodiment can first acquire the current driving basic information (including first vehicle state information, first engine state information, driver driving behavior information, and map data information around the vehicle), then process the current driving basic information based on a pre-trained driver behavior prediction network model to obtain driving behavior information for each future time within a preset future time period starting from the current time, then acquire the current second vehicle state information as the initial state of the vehicle model, and acquire the current second engine state information as the initial state of the engine model, and finally input the driving behavior information for each future time within the preset future time period into the vehicle model and engine model, and use the model predictive control algorithm to calculate the engine thermal management control quantity in the next control cycle with the minimum fuel quantity within the preset future time period as the optimization objective. Therefore, the embodiments of this application not only aim to reduce fuel consumption and improve fuel economy, but also automatically predict the engine thermal management control quantity in the next control cycle to better improve the impact of engine thermal management control on fuel economy. Furthermore, when predicting future driving behavior information, the influence of map data information around the vehicle on engine operating conditions, such as vehicle climbing conditions, turning conditions, intersection deceleration conditions, and speed limit conditions, can be considered. Thus, regardless of how the operating conditions change, the fuel consumption can be minimized, further improving the impact of engine thermal management control on fuel economy.

[0094] Based on the above embodiments, another embodiment of this application also provides an engine thermal management control system, such as... Figure 7 As shown, the system includes a driver behavior prediction network model 510, a vehicle model 520, and an engine model 530; The driver behavior prediction network model 510 is used to acquire the basic driving information at the current moment and process the basic driving information at the current moment to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on the basic driving information at multiple historical moments. The basic driving information includes first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle. The vehicle model 520 is used to obtain the current state information of the second vehicle as the initial state. The engine model 530 is used to obtain the current state information of the second engine as the initial state, and combined with the initialized vehicle model, with the minimum fuel quantity within the preset future time period as the optimization objective, and using the model predictive control algorithm and the driving behavior information of each future time within the preset future time period, calculates the engine thermal management control quantity in the next control cycle.

[0095] In one possible implementation, the first vehicle state information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position; and / or, The second vehicle status information includes at least one of vehicle speed, vehicle pose information, heading angle, and yaw rate; and / or, The second engine status information includes at least one of the following: core rotational dynamics status, intake and exhaust system status, torque and combustion status, and thermodynamics and aftertreatment system status.

[0096] In one possible implementation, such as Figure 5 As shown, the engine model includes a thermal management module, an external characteristic model, an engine management system model, and a combustion model. The thermal management module includes a cooling system model and a thermal management actuator model. The vehicle model is used to obtain the brake pedal opening, steering wheel angle and gear position at each future time, and to simulate and calculate the engine speed at the i-th future time based on the engine torque, brake pedal opening, steering wheel angle and gear position at the i-th future time, and transmit it to the external characteristic model. The external characteristic model is used to transmit the engine speed at the i-th future moment to the thermal management actuator model; The thermal management actuator model is used to calculate the internal circulation coolant flow rate, the external circulation coolant flow rate, and the radiator air flow rate at the i-th future time based on the thermal management control quantity calculated at the i-th future time during the process of minimizing fuel quantity using the model predictive control algorithm, and the engine speed at the i-th future time, and transmit these three flow rates to the cooling system model. The engine management system model is used to calculate the fuel injection quantity at the i-th future time based on the coolant temperature at the i-th future time and transmit it to the combustion model. The combustion model is used to calculate the engine combustion heat flow at the (i+1)th future time based on the fuel injection quantity and the oil temperature at the i-th future time. The cooling system model is used to calculate the coolant temperature and oil temperature at the (i+1)th future time based on the engine combustion heat flow, the internal circulation coolant flow, the external circulation coolant flow, and the radiator air flow at the (i+1)th future time, and transmit them to the engine management system model and the combustion model, respectively. The calculation is iterated in this way until the fuel injection quantity for the preset future time is minimized, and the corresponding thermal management control quantity is obtained.

[0097] In one possible implementation, the external characteristic model is used to obtain the accelerator pedal opening at each future moment within the preset future time period; The vehicle model is used to simulate and calculate the engine speed at the (i-1)th future time and transmit it to the external characteristic model, and to simulate and calculate the vehicle speed at the (i-1)th future time and transmit it to the thermal management actuator model, based on the engine torque at the (i-1)th future time, the brake pedal opening at the (i-1)th future time, the steering wheel angle at the (i-1)th future time, and the gear position at the (i-1)th future time. The engine torque at the first future time is determined based on the initial state of the vehicle model. The external characteristic model is used to send the engine speed at the (i-1)th future time step to the thermal management actuator model; The thermal management actuator model is used to calculate the additional power at the (i-1)th future time based on the thermal management control quantity, engine speed, and vehicle speed calculated by the model predictive control algorithm during the process of minimizing fuel quantity. The external characteristic model is also used to calculate the engine torque at the i-th future time based on the accelerator pedal opening at the (i-1)-th future time, the engine speed at the (i-1)-th future time, and the additional power at the (i-1)-th future time.

[0098] The engine thermal management control system provided in this application embodiment can first obtain the current driving basic information (including first vehicle state information, first engine state information, driver driving behavior information, and map data information around the vehicle) from a pre-trained driver behavior prediction network model, and process the current driving basic information to obtain driving behavior information for each future time within a preset future time period starting from the current time. Then, the vehicle model obtains the current second vehicle state information as the initial state, and the engine model obtains the current second engine state information as the initial state. Finally, the engine model, combined with the vehicle model, uses the model predictive control algorithm and the driving behavior information for each future time within the preset future time period as the optimization objective to calculate the engine thermal management control quantity for the next control cycle. Therefore, the embodiments of this application not only aim to reduce fuel consumption and improve fuel economy, but also automatically predict the engine thermal management control quantity in the next control cycle to better improve the impact of engine thermal management control on fuel economy. Furthermore, when predicting future driving behavior information, the influence of map data information around the vehicle on engine operating conditions, such as vehicle climbing conditions, turning conditions, intersection deceleration conditions, and speed limit conditions, can be considered. Thus, regardless of how the operating conditions change, the fuel consumption can be minimized, further improving the impact of engine thermal management control on fuel economy.

[0099] Based on the above method embodiments, another embodiment of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in any of the above method embodiments.

[0100] Based on the above method embodiments, another embodiment of this application provides an electronic device, the electronic device comprising: One or more processors; Storage device for storing one or more programs. When the one or more programs are executed by the one or more processors, the one or more processors cause the one or more processors to implement the method as described in any of the above method embodiments.

[0101] Based on the above method embodiments, another embodiment of this application provides a computer program product, which includes instructions that, when executed on a computer or processor, cause the computer or processor to perform the method described in any of the above method embodiments.

[0102] The above-described virtual device embodiments and physical device embodiments correspond to the method embodiments and have the same technical effects. For detailed descriptions, please refer to the method embodiments. The device embodiments are derived based on the method embodiments; detailed descriptions can be found in the method embodiments section, and will not be repeated here. Those skilled in the art will understand that the accompanying drawings are merely schematic diagrams of one embodiment, and the modules or processes shown in the drawings are not necessarily essential for implementing this application.

[0103] Those skilled in the art will understand that the modules in the apparatus of the embodiments can be distributed in the apparatus of the embodiments as described in the embodiments, or they can be located in one or more devices different from this embodiment with corresponding changes. The modules of the above embodiments can be combined into one module, or they can be further divided into multiple sub-modules.

[0104] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended 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 of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. An engine thermal management control method, characterized in that, The method includes: Obtain basic driving information at the current moment, wherein the basic driving information includes first vehicle status information, first engine status information, driver's driving behavior information, and map data information around the vehicle; The driver behavior prediction network model is used to process the basic driving information at the current moment to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on the basic driving information at multiple historical moments. The current state information of the second vehicle is used as the initial state of the vehicle model, and the current state information of the second engine is used as the initial state of the engine model. After inputting the driving behavior information of each future moment within the preset future time period into the vehicle model and the engine model, the engine thermal management control quantity in the next control cycle is calculated using the model predictive control algorithm with the minimum fuel quantity within the preset future time period as the optimization objective.

2. The method according to claim 1, characterized in that, The training method for the driver behavior prediction network model includes: Obtain a training set, wherein the training set includes basic driving information at multiple historical moments, and ground truth values ​​of driving behavior information at each moment within the preset future duration starting from each of the historical moments; The initial driver behavior prediction network model is trained using the training set, and the predicted value of driving behavior information is output for each time within the preset future time period starting from each historical time. The current loss value is calculated based on the difference between the true value of the driving behavior information and the predicted value of the driving behavior information at the same moment. The model parameters of the initial driver behavior prediction network model are adjusted using the current loss value, and the driver behavior prediction network model with adjusted parameters is trained until the preset convergence condition is met, thus obtaining the final driver behavior prediction network model.

3. The method according to claim 2, characterized in that, When the driving behavior information includes multiple driving control behaviors, the true value of the driving behavior information includes multiple true values ​​of driving control behaviors, and the predicted value of the driving behavior information includes multiple predicted values ​​of driving control behaviors. Based on the difference between the true value and the predicted value of the driving behavior information at the same moment, the current loss value is calculated, including: For the true value and predicted value of the driving behavior information at the same moment, calculate the difference between the true value of each driving control behavior and its corresponding predicted value, and calculate the behavior prediction loss value at the same moment based on each difference and the driving control behavior weight corresponding to each difference. The current loss value is calculated based on the time weight of each moment within the preset future time period and the predicted loss value of each moment.

4. The method according to any one of claims 1-3, characterized in that, The first vehicle status information includes at least one of vehicle speed, acceleration, sideslip angle, and yaw rate; and / or, The first engine status information includes at least one of the following: engine speed, engine power, engine inlet water temperature, engine outlet water temperature, radiator outlet water temperature, and fuel injection quantity; and / or, The map data information includes at least one of the following within a preset distance ahead of the vehicle: road elevation, road curvature, traffic lights, traffic signs, and whether there is an intersection; and / or, The driving behavior information includes at least one of the following: accelerator pedal opening, brake pedal opening, steering wheel angle, and gear position; and / or, The second vehicle status information includes at least one of vehicle speed, vehicle pose information, heading angle, and yaw rate; and / or, The second engine status information includes at least one of the following: core rotational dynamics status, intake and exhaust system status, torque and combustion status, and thermodynamics and aftertreatment system status.

5. The method according to claim 4, characterized in that, The engine model includes a thermal management module, an external characteristic model, an engine management system model, and a combustion model. The thermal management module includes a cooling system model and a thermal management actuator model. After inputting the driving behavior information for each future moment within the preset future time period into the vehicle model and the engine model, and using the minimum fuel consumption within the preset future time period as the optimization objective, the engine thermal management control quantity for the next control cycle is calculated using a model predictive control algorithm, including: The brake pedal opening, steering wheel angle, and gear position at each future moment are input into the vehicle model; The vehicle model simulates and calculates the engine speed at the i-th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the i-th future time, and transmits this information to the external characteristic model so that the external characteristic model can transmit the engine speed at the i-th future time to the thermal management actuator model. The thermal management actuator model calculates the internal circulation coolant flow rate, the external circulation coolant flow rate, and the radiator air flow rate at the i-th future time based on the thermal management control quantity calculated by the model predictive control algorithm during the process of minimizing fuel quantity, and transmits these three flow rates to the cooling system model. The engine management system model calculates the fuel injection quantity for the i-th future time based on the coolant temperature at the i-th future time and transmits it to the combustion model. The combustion heat flow of the engine at the (i+1)th future moment is calculated using the combustion model based on the fuel injection quantity and oil temperature at the i-th future moment. The cooling system model calculates the coolant temperature and oil temperature at the (i+1)th future time based on the engine combustion heat flow, the internal circulation coolant flow, the external circulation coolant flow, and the radiator air flow at the (i+1)th future time. These values ​​are then transmitted to the engine management system model and the combustion model, respectively. This process is iteratively repeated until the calculated fuel injection amount for the preset future time is minimized, at which point the corresponding thermal management control value is obtained.

6. The method according to claim 5, characterized in that, After inputting the driving behavior information for each future moment within the preset future time period into the vehicle model and the engine model, and using the minimum fuel consumption within the preset future time period as the optimization objective, the engine thermal management control quantity for the next control cycle is calculated using a model predictive control algorithm. The method also includes: The accelerator pedal opening at each future moment within the preset future time period is input into the external characteristic model; The method for determining the engine torque at the i-th future moment includes: The vehicle model simulates and calculates the engine speed at the (i-1)th future time based on the engine torque, brake pedal opening, steering wheel angle, and gear position at the (i-1)th future time, and transmits it to the external characteristic model. It also simulates and calculates the vehicle speed at the (i-1)th future time and transmits it to the thermal management actuator model. The engine torque at the first future time is determined based on the initial state of the vehicle model. The external characteristic model sends the engine speed at the (i-1)th future time to the thermal management actuator model. After the thermal management actuator model calculates the additional power at the (i-1)th future time based on the thermal management control quantity, engine speed, and vehicle speed calculated by the model predictive control algorithm during the process of minimizing fuel quantity, the external characteristic model calculates the engine torque at the (i-1)th future time based on the accelerator pedal opening, engine speed, and additional power at the (i-1)th future time.

7. An engine thermal management control device, characterized in that, The device includes: The acquisition unit is used to acquire the basic driving information at the current moment, wherein the basic driving information includes first vehicle status information, first engine status information, driver's driving behavior information, and map data information around the vehicle. The prediction unit is used to process the basic driving information at the current moment based on the driver behavior prediction network model to obtain the driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on the basic driving information at multiple historical moments. An initialization unit is used to obtain the current state information of the second vehicle as the initial state of the vehicle model, and to obtain the current state information of the second engine as the initial state of the engine model. The calculation unit is used to input the driving behavior information of each future moment within the preset future time period into the vehicle model and the engine model, and then use the model predictive control algorithm to calculate the engine thermal management control quantity in the next control cycle with the minimum fuel quantity within the preset future time period as the optimization objective.

8. An engine thermal management control system, characterized in that, The system includes a driver behavior prediction network model, a vehicle model, and an engine model; The driver behavior prediction network model is used to acquire basic driving information at the current moment and process the basic driving information at the current moment to obtain driving behavior information at each future moment within a preset future time period starting from the current moment. The driver behavior prediction network model is a neural network model trained based on driving basic information at multiple historical moments. The basic driving information includes first vehicle state information, first engine state information, driver's driving behavior information, and map data information around the vehicle. The vehicle model is used to obtain the current state information of the second vehicle as the initial state; The engine model is used to obtain the current state information of the second engine as the initial state, and combined with the initialized vehicle model, with the minimum fuel quantity within the preset future time period as the optimization objective, and using the model predictive control algorithm and the driving behavior information of each future time within the preset future time period, to calculate the engine thermal management control quantity in the next control cycle.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the method as described in any one of claims 1-6.

10. An electronic device, characterized in that, The electronic device includes: One or more processors; Storage device for storing one or more programs. Wherein, when the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-6.

Images