Energy control method and device for hybrid electric vehicle based on driving condition prediction

Abstract

The present disclosure relates to a method and device for controlling the energy of a hybrid electric vehicle based on driving condition prediction. The method comprises: analyzing the power characteristics of the hybrid electric vehicle to generate a power characteristic curve of the hybrid electric vehicle; based on a preset LSTM neural network model, taking the current speed and current power of the hybrid electric vehicle as input, predicting the driving condition to generate a predicted speed and predicted power of the hybrid electric vehicle; generating a predicted vehicle condition based on the predicted speed, predicted power, and power characteristic curve, and generating a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle. The present disclosure achieves the prediction of the driving condition of a hybrid electric vehicle, and performs energy flow control in advance based on the prediction results, thereby reducing the frequency of the engine starting and entering the forced charging mode due to reduced power, and improving energy utilization efficiency.

Description

Technical Field

[0001] The present disclosure relates to the fields of new energy and electric vehicles, and in particular to a method and device for controlling energy of a hybrid electric vehicle based on driving condition prediction. Background Art

[0002] The powertrain of a plug-in hybrid electric vehicle (PHEV) includes not only a fuel engine but also components such as a drive motor and a power battery. Energy control in a HEV involves rationally coordinating and controlling the flow of energy between the various powertrain components. The fuel savings of HEVs come from energy recovery, eliminating engine idling, and operating the engine in its high-efficiency range. However, achieving even higher fuel savings requires a sound energy control strategy. Because each conversion between mechanical and electrical energy results in efficiency losses, a sound energy control strategy for hybrid vehicles with an engine as a driving force should minimize the number of conversions between mechanical and electrical energy.

[0003] The energy control strategy for existing hybrid electric vehicles requires setting the switching timings for various energy flow modes based on the characteristic parameters of the engine, drive motor, and power battery, and within certain thresholds. Once the energy control strategy is set, it remains unchanged. This energy control strategy lacks generalizability and cannot be applied to all vehicle driving conditions. One example is an energy control method for a single-motor plug-in hybrid vehicle, which uses thresholds to differentiate between driving modes during the power consumption phase, the power maintenance phase, and the power replenishment phase. This energy control strategy lacks predictive power for the vehicle's driving conditions. If the power level remains low and is followed by a prolonged period of high-power driving, the vehicle will be unable to quickly replenish its power and will be forced to stop for static recharging before continuing. Energy control methods based on driving conditions rely on a vehicle-road-cloud system to pre-calculate the vehicle's driving conditions. By extracting key features and performing data clustering, the corresponding driving condition category is identified and the corresponding energy control strategy is implemented. However, it is impossible to accurately define the operating condition information by relying solely on the extracted main features (including running time, maximum speed, maximum acceleration, average acceleration, and average deceleration). In addition, the road conditions during vehicle driving are ever-changing, and the above-mentioned features are obviously difficult to accurately predict in advance before the vehicle drives.

[0004] Therefore, one or more methods are needed to solve the above problems.

[0005] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present disclosure, and therefore may include information that does not constitute prior art known to ordinary technicians in the field. Summary of the Invention

[0006] The purpose of the present disclosure is to provide a hybrid electric vehicle energy control method and device based on driving condition prediction, thereby overcoming one or more problems caused by the limitations and defects of related technologies to at least a certain extent.

[0007] According to one aspect of the present disclosure, a method for controlling energy of a hybrid electric vehicle based on driving condition prediction is provided, comprising:

[0008] Analyzing the power characteristics of the hybrid electric vehicle and generating a power characteristic curve of the hybrid electric vehicle in a vehicle speed-power coordinate system;

[0009] Based on a preset LSTM neural network model, the current speed and current power of the hybrid electric vehicle are used as input to predict the driving condition of the hybrid electric vehicle, and generate a predicted speed and predicted power of the hybrid electric vehicle;

[0010] A predicted vehicle condition is generated based on the predicted vehicle speed, predicted power, and power characteristic curve, and a vehicle control instruction is generated based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle.

[0011] In an exemplary embodiment of the present disclosure, the method further includes:

[0012] The series-parallel connection status of the engine and drive motor of the hybrid electric vehicle and the power characteristics during combined operation are analyzed, and the power characteristic curve of the hybrid electric vehicle and the charging and discharging status of the electric vehicle corresponding to the power characteristic curve are generated in the vehicle speed-power coordinate system. The power characteristic curve includes the target torque curve, the external characteristic power curve of the drive motor, the external characteristic power curve of the engine, the lower limit curve of the engine economic operating output power, and the power curve when the engine and the drive motor work in parallel.

[0013] In an exemplary embodiment of the present disclosure, the method further includes training the LSTM neural network model based on historical operating parameters of the hybrid electric vehicle to generate the preset LSTM neural network model:

[0014] Based on the preset hyperparameters of the LSTM neural network model, the historical speed and historical power of the hybrid electric vehicle are used as inputs, and the training of the LSTM neural network model is completed according to the loss function.

[0015] In an exemplary embodiment of the present disclosure, the method further includes:

[0016] Establishing an LSTM neural network model population comprising a preset number of LSTM neural network models, and setting a first random hyperparameter for each LSTM neural network model in the LSTM neural network model population;

[0017] Based on the random hyperparameters set for each of the LSTM neural network models, taking the historical speed and historical power of the hybrid electric vehicle as input, and determining according to the loss function, performing a first round of training on the LSTM neural network models in the LSTM neural network model population to generate first hyperparameter values;

[0018] Calculating the Euclidean distance of the first hyperparameter value of each of the LSTM neural network models, generating a second hyperparameter value based on the Euclidean distance, and generating a second random hyperparameter based on the second hyperparameter value and a random scaling factor;

[0019] A second random hyperparameter is set based on each of the LSTM neural network models, and the historical speed and historical power of the hybrid electric vehicle are used as input. According to the loss function, a second round of training is performed on the LSTM neural network models in the LSTM neural network model population.

[0020] In an exemplary embodiment of the present disclosure, the method further includes:

[0021] Each LSTM neural network model in the LSTM neural network model population is trained for a preset number of rounds based on an iterative hyperparameter method to generate a preset LSTM neural network model.

[0022] In an exemplary embodiment of the present disclosure, the energy control strategy in the method includes:

[0023] When the current vehicle condition is pure electric, and the predicted vehicle condition is pure electric, the energy control strategy is to start the engine to charge the power battery when the power level is lower than the threshold;

[0024] If the current vehicle condition is pure electric, and the predicted vehicle condition is pure oil, the energy control strategy is to do nothing.

[0025] When the vehicle is currently in pure electric mode and is predicted to enter parallel mode, the energy control strategy is to start the engine in advance to charge the power battery.

[0026] When the current vehicle condition is pure oil condition and the predicted vehicle condition is pure oil condition, the energy control strategy is no processing;

[0027] When the vehicle is currently operating in a pure fuel mode and is predicted to enter a parallel mode, the energy control strategy uses a preset portion of the engine power for charging to reserve energy in advance.

[0028] If the vehicle is currently operating in a purely fuel-only mode and is predicted to be operating in a purely electric mode, the energy control strategy uses a preset portion of the engine power for charging to reserve energy in advance.

[0029] When the current vehicle condition is parallel operation and the predicted vehicle condition is parallel operation, the energy control strategy is to use part of the engine power for charging to prevent the power battery from being consumed for a long time and causing the battery power to decrease;

[0030] When the current vehicle condition is parallel operation and the predicted vehicle condition is series operation, the energy control strategy is to do nothing.

[0031] When the vehicle is currently in parallel operation and is predicted to enter pure electric operation, the energy control strategy uses a preset portion of the engine power for charging to prevent prolonged consumption of the power battery, which could lead to a decrease in battery power.

[0032] When the current vehicle condition is parallel operation, the predicted vehicle condition is pure oil operation, and the energy control strategy is no processing;

[0033] When the current vehicle condition is a series condition and the predicted vehicle condition is a series condition, the energy control strategy is to enter the power following mode. The power output of the engine follows the power demand of the drive motor. The power battery is not charged and does not need to be discharged.

[0034] When the current vehicle condition is series operation and the predicted vehicle condition is parallel operation, the energy control strategy is to do nothing.

[0035] When the current vehicle condition is a series operating condition, the predicted vehicle condition is to enter a pure electric operating condition, and the energy control strategy is to do nothing.

[0036] In one aspect of the present disclosure, a hybrid electric vehicle energy control device based on driving condition prediction is provided, comprising:

[0037] A power characteristic curve generating module is used to analyze the power characteristics of the hybrid electric vehicle and generate a power characteristic curve of the hybrid electric vehicle in a vehicle speed-power coordinate system;

[0038] a vehicle condition prediction module, configured to predict the driving condition of the hybrid electric vehicle based on a preset LSTM neural network model and taking the current speed and current power of the hybrid electric vehicle as input, and generate a predicted speed and predicted power of the hybrid electric vehicle;

[0039] The energy control module is used to generate a predicted vehicle condition based on the predicted vehicle speed, predicted power, and power characteristic curve, and to generate a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle.

[0040] In an exemplary embodiment of the present disclosure, a hybrid electric vehicle energy control method based on driving condition prediction is disclosed, wherein the method includes: analyzing the power characteristics of the hybrid electric vehicle to generate a power characteristic curve of the hybrid electric vehicle; based on a preset LSTM neural network model, using the current speed and current power of the hybrid electric vehicle as input, predicting the driving condition to generate a predicted speed and predicted power of the hybrid electric vehicle; generating a predicted vehicle condition based on the predicted speed, predicted power, and power characteristic curve, and generating a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle. The present disclosure achieves the prediction of the driving condition of the hybrid electric vehicle and performs energy flow control in advance based on the prediction results, reducing the frequency of the engine starting and entering the forced charging mode due to reduced power, thereby improving energy utilization efficiency.

[0041] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The above and other features and advantages of the present disclosure will become more apparent by describing in detail example embodiments thereof with reference to the accompanying drawings.

[0043] Figure 1 A flowchart of a hybrid electric vehicle energy control method based on driving condition prediction according to an exemplary embodiment of the present disclosure is shown;

[0044] Figure 2 A schematic diagram of a power characteristic curve of a hybrid electric vehicle energy control method based on driving condition prediction according to an exemplary embodiment of the present disclosure is shown;

[0045] Figure 3 A schematic diagram showing a predicted vehicle condition of a hybrid electric vehicle energy control method based on driving condition prediction according to an exemplary embodiment of the present disclosure is shown;

[0046] Figure 4 A schematic block diagram of an energy control device for a hybrid electric vehicle based on driving condition prediction according to an exemplary embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0047] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be embodied in many forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete and will fully convey the concepts of the example embodiments to those skilled in the art. Like reference numerals in the drawings represent like or similar parts, and thus repetitive description thereof will be omitted.

[0048] In addition, the described features, structures or characteristics may be combined in any suitable manner in one or more embodiments. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present disclosure. However, those skilled in the art will appreciate that the technical solutions of the present disclosure can be practiced without one or more of the specific details, or other methods, components, materials, devices, steps, etc. can be adopted. In other cases, well-known structures, methods, devices, implementations, materials or operations are not shown or described in detail to avoid obscuring various aspects of the present disclosure.

[0049] The blocks shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically separate entities. Specifically, these functional entities may be implemented in software, or in one or more software-hardened modules, or in different networks and / or processor devices and / or microcontroller devices.

[0050] In this exemplary embodiment, a hybrid electric vehicle energy control method based on driving condition prediction is first provided; Figure 1 As shown in , the energy control method of a hybrid electric vehicle based on driving condition prediction may include the following steps:

[0051] Step S110 , analyzing the power characteristics of the hybrid electric vehicle and generating a power characteristic curve of the hybrid electric vehicle in a vehicle speed-power coordinate system;

[0052] Step S120, based on a preset LSTM neural network model, using the current speed and current power of the hybrid electric vehicle as input, predicting the driving condition of the hybrid electric vehicle to generate a predicted speed and predicted power of the hybrid electric vehicle;

[0053] Step S130 , generating a predicted vehicle condition based on the predicted vehicle speed, predicted power, and power characteristic curve, and generating a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle.

[0054] In an exemplary embodiment of the present disclosure, a hybrid electric vehicle energy control method based on driving condition prediction is disclosed, wherein the method includes: analyzing the power characteristics of the hybrid electric vehicle to generate a power characteristic curve of the hybrid electric vehicle; based on a preset LSTM neural network model, using the current speed and current power of the hybrid electric vehicle as input, predicting the driving condition to generate a predicted speed and predicted power of the hybrid electric vehicle; generating a predicted vehicle condition based on the predicted speed, predicted power, and power characteristic curve, and generating a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle. The present disclosure achieves the prediction of the driving condition of the hybrid electric vehicle and performs energy flow control in advance based on the prediction results, reducing the frequency of the engine starting and entering the forced charging mode due to reduced power, thereby improving energy utilization efficiency.

[0055] Next, a hybrid electric vehicle energy control method based on driving condition prediction in this example embodiment will be further described.

[0056] In step S110 , the power characteristics of the hybrid electric vehicle may be analyzed, and a power characteristic curve of the hybrid electric vehicle may be generated in a vehicle speed-power coordinate system.

[0057] In this exemplary embodiment, the method further includes:

[0058] The series-parallel connection status of the engine and drive motor of the hybrid electric vehicle and the power characteristics during combined operation are analyzed, and the power characteristic curve of the hybrid electric vehicle and the charging and discharging status of the electric vehicle corresponding to the power characteristic curve are generated in the vehicle speed-power coordinate system. The power characteristic curve includes the target torque curve, the external characteristic power curve of the drive motor, the external characteristic power curve of the engine, the lower limit curve of the engine economic operating output power, and the power curve when the engine and the drive motor work in parallel.

[0059] In this exemplary embodiment, Figure 2 As shown in FIG. , it is the power characteristic curve of the hybrid electric vehicle, wherein:

[0060] 1. C represents the speed at which the engine can drive the vehicle, and I represents the maximum design speed of the vehicle.

[0061] 2. Curve BHN represents the target torque and the required drive power when the driver depresses the accelerator pedal 100% at the corresponding vehicle speed. This drive power can be provided by the engine and drive motor in series or in parallel.

[0062] 3. Curve AFL represents the curve of the external characteristic power of the drive motor changing with vehicle speed;

[0063] 4. Curve EK represents the variation of the engine's external characteristic power with vehicle speed. In the hybrid architecture of this embodiment of the present invention, the engine engages in direct drive only after the vehicle speed exceeds 40 km / h. Therefore, only the power characteristics above 40 km / h are captured.

[0064] 5. Curve DJ represents the lower limit of the engine's output power in the high-efficiency zone. Below this power, the engine's specific fuel consumption increases sharply.

[0065] 6. Curve GM represents the upper limit of the total power that can be achieved when the engine and drive motor are driven in parallel, due to the limitation of the continuous discharge power of the power battery;

[0066] 7. The areas enclosed by each curve are defined as series mode, parallel mode, pure electric mode, or pure gasoline mode. Pure electric mode and parallel mode both cause the power battery charge to decrease, hence the "↓" symbol. In series mode and pure gasoline mode, the engine starts, allowing for optional battery charging, hence the "↑" symbol.

[0067] When a hybrid electric vehicle can consistently operate within its power characteristics, any drop in battery charge can be balanced by the added charge, eliminating the need to start the engine for forced recharging. However, under certain operating conditions, a continued drop in battery charge may require the vehicle to remain stationary for an extended period of time.

[0068] The present invention aims to make the hybrid electric vehicle always run within the power characteristic diagram by predicting the driving conditions, reduce the frequency of starting the engine to enter the forced power replenishment mode due to insufficient power, and replenish the power of the power battery in advance during the operation of the engine.

[0069] In step S120, based on a preset LSTM neural network model, the current speed and current power of the hybrid electric vehicle are used as input to predict the driving condition of the hybrid electric vehicle and generate the predicted speed and predicted power of the hybrid electric vehicle.

[0070] In the embodiment of this example, the method further includes training the LSTM neural network model based on historical operating parameters of the hybrid electric vehicle to generate the preset LSTM neural network model:

[0071] Based on the preset hyperparameters of the LSTM neural network model, the historical speed and historical power of the hybrid electric vehicle are used as inputs, and the training of the LSTM neural network model is completed according to the loss function.

[0072] In this exemplary embodiment, the method further includes:

[0073] Establishing an LSTM neural network model population comprising a preset number of LSTM neural network models, and setting a first random hyperparameter for each LSTM neural network model in the LSTM neural network model population;

[0074] Based on the random hyperparameters set for each of the LSTM neural network models, taking the historical speed and historical power of the hybrid electric vehicle as input, and determining according to the loss function, performing a first round of training on the LSTM neural network models in the LSTM neural network model population to generate first hyperparameter values;

[0075] Calculating the Euclidean distance of the first hyperparameter value of each of the LSTM neural network models, generating a second hyperparameter value based on the Euclidean distance, and generating a second random hyperparameter based on the second hyperparameter value and a random scaling factor;

[0076] A second random hyperparameter is set based on each of the LSTM neural network models, and the historical speed and historical power of the hybrid electric vehicle are used as input. According to the loss function, a second round of training is performed on the LSTM neural network models in the LSTM neural network model population.

[0077] In this exemplary embodiment, the method further includes:

[0078] Each LSTM neural network model in the LSTM neural network model population is trained for a preset number of rounds based on an iterative hyperparameter method to generate a preset LSTM neural network model.

[0079] In the embodiment of this example, the present invention aims to make the hybrid electric vehicle always run within the power characteristic diagram by predicting the driving conditions, reduce the frequency of starting the engine to enter the forced power replenishment mode due to insufficient power, and replenish the power of the power battery in advance during the operation of the engine.

[0080] In this example, the C-WTVC operating condition is used as an example. This condition defines a curve showing the change in vehicle speed over time. In fact, if the vehicle's sliding resistance coefficient is known, a curve showing the change in (vehicle speed, power) over time can be calculated based on the curve showing the change in vehicle speed over time. This curve is plotted as a three-dimensional spatial curve, representing a time series (a variable that changes over time). LSTM can be used to learn and predict time series. Based on the current (vehicle speed, power) parameters and past (vehicle speed, power) change trends, LSTM can achieve the goal of predicting future time series through a certain degree of discard learning.

[0081] In this exemplary embodiment, Figure 3The three-dimensional power characteristic graph shows how (vehicle speed and power) change over time. The thick 3D curve on the left represents the historical information input to the LSTM network, while the thin 3D curve on the right represents the driving condition information predicted by the LSTM network. During this period, the vehicle was operating under pure fuel conditions.

[0082] To achieve accurate prediction, we need to train the LSTM neural network to learn how vehicle speed and power change over time so that it can make accurate predictions.

[0083] Therefore, we use the following population optimization method to train a group of LSTM neural networks in order to identify the most effective one. During training, to improve LSTM prediction accuracy, we build n LSTM networks, each with randomly set initial hyperparameters. This is like building a population, and we execute our population optimization algorithm using the following scheme. Hyperparameters are preset values for a neural network before training, including the number of hidden layers, the number of training cycles, the batch size, the dropout rate, the learning rate, and the gradient threshold. Different hyperparameter settings can result in different trained LSTM neural networks.

[0084] Assume that the initial hyperparameters of the i-th LSTM network are set to Each of these The initial value is a random number. Define the mean square error (MSE) as the loss function. After each training, observe and compare the loss function of each LSTM network, and select the LSTM hyperparameter value with the smallest loss function. As the direction of the next round of training optimization. Before the next round of training begins, update the hyperparameters of all LSTM networks. The specific update method is to calculate the hyperparameter vector of each LSTM network and the optimal hyperparameter vector The Euclidean distance between the two is calculated by adding the random scaling factor η and the Levy motion ψ to calculate the hyperparameter value for the next round of training. The calculation formula is:

[0085]

[0086] The left side of the above formula represents the hyperparameter vector of the j+1th iteration of the i-th LSTM network. The first term on the right side is the hyperparameter vector of the j-th iteration of the i-th LSTM network. The second term on the right side is the Euclidean distance between the hyperparameter vector of the j-th iteration of the i-th LSTM network and the hyperparameter vector of the optimal LSTM network at the j-th iteration. The third term on the right side is the Levy motion, which adds random motion to prevent the neural network training process from falling into a local optimal solution.

[0087] After several iterations of optimization, we find the LSTM network with the smallest loss function, which is the best result of the population optimization algorithm.

[0088] LSTM training input can be derived from experimentally measured vehicle driving data or from driving data recorded by the vehicle-road-cloud system for previously sold vehicles. In short, diverse data sources can meet the data diversity requirements of neural network training.

[0089] By recording real-time changes in (vehicle speed and power) over time, a trained LSTM neural network can predict the likely future trends in (vehicle speed and time). If (vehicle speed and time) are concentrated in the power characteristic graph's declining battery area over a period of time, the power battery can be replenished economically in advance to prevent sudden power shortages. A slow, high-power, rapid replenishment strategy would be necessary to mitigate fuel consumption, potentially preventing the engine from replenishing power at an economical speed.

[0090] In step S130, a predicted vehicle condition may be generated based on the predicted vehicle speed, predicted power, and power characteristic curve, and a vehicle control instruction may be generated based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle.

[0091] In this exemplary embodiment, the energy control strategy in the method includes:

[0092] When the current vehicle condition is pure electric, and the predicted vehicle condition is pure electric, the energy control strategy is to start the engine to charge the power battery when the power level is lower than the threshold;

[0093] If the current vehicle condition is pure electric, and the predicted vehicle condition is pure oil, the energy control strategy is to do nothing.

[0094] When the vehicle is currently in pure electric mode and is predicted to enter parallel mode, the energy control strategy is to start the engine in advance to charge the power battery.

[0095] When the current vehicle condition is pure oil condition and the predicted vehicle condition is pure oil condition, the energy control strategy is no processing;

[0096] When the vehicle is currently operating in a pure fuel mode and is predicted to enter a parallel mode, the energy control strategy uses a preset portion of the engine power for charging to reserve energy in advance.

[0097] If the vehicle is currently operating in a purely fuel-only mode and is predicted to be operating in a purely electric mode, the energy control strategy uses a preset portion of the engine power for charging to reserve energy in advance.

[0098] When the current vehicle condition is parallel operation and the predicted vehicle condition is parallel operation, the energy control strategy is to use part of the engine power for charging to prevent the power battery from being consumed for a long time and causing the battery power to decrease;

[0099] When the current vehicle condition is parallel operation and the predicted vehicle condition is series operation, the energy control strategy is to do nothing.

[0100] When the vehicle is currently in parallel operation and is predicted to enter pure electric operation, the energy control strategy uses a preset portion of the engine power for charging to prevent prolonged consumption of the power battery, which could lead to a decrease in battery power.

[0101] When the current vehicle condition is parallel operation, the predicted vehicle condition is pure oil operation, and the energy control strategy is no processing;

[0102] When the current vehicle condition is a series condition and the predicted vehicle condition is a series condition, the energy control strategy is to enter the power following mode. The power output of the engine follows the power demand of the drive motor. The power battery is not charged and does not need to be discharged.

[0103] When the current vehicle condition is series operation and the predicted vehicle condition is parallel operation, the energy control strategy is to do nothing.

[0104] When the current vehicle condition is a series operating condition, the predicted vehicle condition is to enter a pure electric operating condition, and the energy control strategy is to do nothing.

[0105] In this example embodiment, after the LSTM network is trained, it can be used to predict the time series of (vehicle speed, power). A fully trained LSTM network can predict the range within which (vehicle speed, power) will fall within a future period based on historically collected (vehicle speed, power) data. If it is predicted that future driving conditions will involve operating in the power characteristic diagram's reduced charge region for a prolonged period at high power, then the battery is intentionally replenished in advance under the current operating conditions. For example, if the vehicle is currently operating in the OAFC region of the power characteristic diagram near the upper bound, the engine is started and enters series mode, operating at an economical rate, with excess power used to charge the power battery. If the vehicle is currently operating in the CDJI region of the power characteristic diagram near the upper bound, the engine is started and enters pure fuel mode, operating at an economical rate, with excess power used to charge the power battery. If the vehicle is currently operating in the DEKJ region of the power characteristic diagram, the engine output power is appropriately increased, without exceeding the engine's external characteristic power, and the excess power is used to charge the power battery. This can offset the negative impact of a continued decrease in charge within a future period of driving conditions.

[0106] Specifically, the energy control strategy can be implemented according to the following table:

[0107]

[0108]

[0109] In this exemplary embodiment, the present invention predicts the driving conditions of hybrid electric vehicles and proactively controls energy flow based on the predicted results, reducing the frequency of engine startups entering forced recharging mode due to battery drain, thereby improving energy efficiency. A trained LSTM network can predict future driving conditions based on current and historical driving condition information. This prediction adapts to complex and changing driving conditions, enabling energy control strategies to adapt to the vehicle's operating conditions and improving the generalizability of the control method.

[0110] It should be noted that although the steps of the method disclosed herein are depicted in a particular order in the accompanying drawings, this does not require or imply that the steps must be performed in that particular order, or that all steps must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one, and / or one step may be decomposed into multiple steps.

[0111] In addition, in this exemplary embodiment, a hybrid electric vehicle energy control device based on driving condition prediction is also provided. Figure 4 As shown, the hybrid electric vehicle energy control device 400 based on driving condition prediction may include: a power characteristic curve generation module 410, a vehicle condition prediction module 420 and an energy control module 430.

[0112] The power characteristic curve generating module 410 is used to analyze the power characteristics of the hybrid electric vehicle and generate a power characteristic curve of the hybrid electric vehicle in a vehicle speed-power coordinate system;

[0113] a vehicle condition prediction module 420 for predicting the driving condition of the hybrid electric vehicle based on a preset LSTM neural network model and taking the current speed and current power of the hybrid electric vehicle as input, and generating a predicted speed and predicted power of the hybrid electric vehicle;

[0114] The energy control module 430 is used to generate a predicted vehicle condition based on the predicted vehicle speed, predicted power, and power characteristic curve, and to generate a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle.

[0115] The specific details of each of the above hybrid electric vehicle energy control device modules based on driving condition prediction have been described in detail in the corresponding hybrid electric vehicle energy control method based on driving condition prediction, and will not be repeated here.

[0116] It should be noted that although the above detailed description mentions several modules or units of a hybrid electric vehicle energy control device 400 based on driving condition prediction, this division is not mandatory. In fact, according to embodiments of the present disclosure, the features and functions of two or more modules or units described above can be embodied in a single module or unit. Conversely, the features and functions of a single module or unit described above can be further divided and embodied by multiple modules or units.

[0117] Furthermore, the above-described figures are merely illustrative of the processes included in the method according to exemplary embodiments of the present invention and are not intended to be limiting. It is readily understood that the processes illustrated in the above-described figures do not indicate or limit the temporal order of these processes. Furthermore, it is readily understood that these processes may be executed synchronously or asynchronously, for example, in multiple modules.

[0118] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow from the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the claims.

[0119] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A hybrid electric vehicle energy control method based on driving condition prediction, characterized in that: The method comprises: Analyzing the power characteristics of the hybrid electric vehicle and generating a power characteristic curve of the hybrid electric vehicle in a vehicle speed-power coordinate system; Based on a preset LSTM neural network model, the current speed and current power of the hybrid electric vehicle are used as input to predict the driving condition of the hybrid electric vehicle, and generate a predicted speed and predicted power of the hybrid electric vehicle; generating a predicted vehicle condition based on the predicted vehicle speed, predicted power, and power characteristic curve, and generating a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle; The energy control strategy in the method includes: When the current vehicle condition is pure electric, the predicted vehicle condition is pure electric, and the energy control strategy is to start the engine to charge the power battery when the power level is lower than the threshold; When the current vehicle condition is pure electric, the predicted vehicle condition is to enter pure oil condition, and the energy control strategy is to do nothing; When the current vehicle condition is pure electric, the vehicle condition is predicted to enter parallel operation, and the energy control strategy is to start the engine in advance to charge the power battery; When the current vehicle condition is pure oil condition, the predicted vehicle condition is pure oil condition, and the energy control strategy is no processing; When the current vehicle condition is pure oil operation, the vehicle condition is predicted to enter parallel operation. The energy control strategy is to preset part of the engine power for charging to reserve electric energy in advance. When the vehicle is currently operating in a purely oil-powered mode and is predicted to enter a purely electric mode, the energy control strategy uses a preset portion of the engine power for charging to reserve energy in advance. When the current vehicle condition is parallel operation, the energy control strategy is to use part of the engine power for charging to prevent the power battery from being consumed for a long time and causing the battery power to decrease; When the current vehicle condition is parallel operation, the predicted vehicle condition is to enter series operation, and the energy control strategy is to do nothing; When the vehicle is in parallel operation and is predicted to enter pure electric operation, the energy control strategy uses a preset portion of the engine power for charging to prevent prolonged consumption of the power battery, which could lead to a decrease in battery power. When the current vehicle condition is parallel operation, the vehicle condition is predicted to enter the pure oil operation condition, and the energy control strategy is to do nothing; When the current vehicle condition is a series operating condition, or the vehicle condition is predicted to be a series operating condition, the energy control strategy is to enter a power following mode, where the power output of the engine follows the power demand of the drive motor, without charging the power battery or discharging the power battery. When the current vehicle condition is a series condition, the vehicle condition is predicted to enter a parallel condition, and the energy control strategy is to do nothing; When the current vehicle condition is a series operating condition, the predicted vehicle condition is to enter a pure electric operating condition, and the energy control strategy is to do nothing.

2. The hybrid electric vehicle energy control method based on driving condition prediction according to claim 1, characterized in that: The method further comprises: The series-parallel connection status of the engine and drive motor of the hybrid electric vehicle and the power characteristics during combined operation are analyzed, and the power characteristic curve of the hybrid electric vehicle and the charging and discharging status of the electric vehicle corresponding to the power characteristic curve are generated in the vehicle speed-power coordinate system. The power characteristic curve includes the target torque curve, the external characteristic power curve of the drive motor, the external characteristic power curve of the engine, the lower limit curve of the engine economic operating output power, and the power curve when the engine and the drive motor work in parallel.

3. The hybrid electric vehicle energy control method based on driving condition prediction according to claim 1, characterized in that: The method further includes training the LSTM neural network model based on historical operating parameters of the hybrid electric vehicle to generate the preset LSTM neural network model: Based on the preset hyperparameters of the LSTM neural network model, the historical speed and historical power of the hybrid electric vehicle are used as inputs, and the training of the LSTM neural network model is completed according to the loss function.

4. The hybrid electric vehicle energy control method based on driving condition prediction according to claim 3, characterized in that: The method further comprises: Establishing an LSTM neural network model population comprising a preset number of LSTM neural network models, and setting a first random hyperparameter for each LSTM neural network model in the LSTM neural network model population; Based on the random hyperparameters set for each of the LSTM neural network models, taking the historical speed and historical power of the hybrid electric vehicle as input, and determining according to the loss function, performing a first round of training on the LSTM neural network models in the LSTM neural network model population to generate first hyperparameter values; Calculating the Euclidean distance of the first hyperparameter value of each of the LSTM neural network models, screening and generating a second hyperparameter value based on the Euclidean distance, and generating a second random hyperparameter based on the second hyperparameter value and a random scaling factor; A second random hyperparameter is set based on each of the LSTM neural network models, and the historical speed and historical power of the hybrid electric vehicle are used as input. According to the loss function, a second round of training is performed on the LSTM neural network models in the LSTM neural network model population.

5. The hybrid electric vehicle energy control method based on driving condition prediction according to claim 4, characterized in that: The method further comprises: Each LSTM neural network model in the LSTM neural network model population is trained for a preset number of rounds based on an iterative hyperparameter method to generate a preset LSTM neural network model.

6. A hybrid electric vehicle energy control device based on driving condition prediction, characterized in that: The device comprises: A power characteristic curve generating module is used to analyze the power characteristics of the hybrid electric vehicle and generate a power characteristic curve of the hybrid electric vehicle in a vehicle speed-power coordinate system; a vehicle condition prediction module, configured to predict the driving condition of the hybrid electric vehicle based on a preset LSTM neural network model and taking the current speed and current power of the hybrid electric vehicle as input, and generate a predicted speed and predicted power of the hybrid electric vehicle; an energy control module, configured to generate a predicted vehicle condition based on the predicted vehicle speed, predicted power, and power characteristic curve, and to generate a vehicle control instruction based on the predicted vehicle condition and a preset energy control strategy to complete energy control of the hybrid electric vehicle; The energy control strategy includes: When the current vehicle condition is pure electric, the predicted vehicle condition is pure electric, and the energy control strategy is to start the engine to charge the power battery when the power level is lower than the threshold; When the current vehicle condition is pure electric, the predicted vehicle condition is to enter pure oil condition, and the energy control strategy is to do nothing; When the current vehicle condition is pure electric, the vehicle condition is predicted to enter parallel operation, and the energy control strategy is to start the engine in advance to charge the power battery; When the current vehicle condition is pure oil condition, the predicted vehicle condition is pure oil condition, and the energy control strategy is no processing; When the current vehicle condition is pure oil operation, the vehicle condition is predicted to enter parallel operation. The energy control strategy is to preset part of the engine power for charging to reserve electric energy in advance. When the vehicle is currently operating in a purely oil-powered mode and is predicted to enter a purely electric mode, the energy control strategy uses a preset portion of the engine power for charging to reserve energy in advance. When the current vehicle condition is parallel operation, the energy control strategy is to use part of the engine power for charging to prevent the power battery from being consumed for a long time and causing the battery power to decrease; When the current vehicle condition is parallel operation, the predicted vehicle condition is to enter series operation, and the energy control strategy is to do nothing; When the vehicle is in parallel operation and is predicted to enter pure electric operation, the energy control strategy uses a preset portion of the engine power for charging to prevent prolonged consumption of the power battery, which could lead to a decrease in battery power. When the current vehicle condition is parallel operation, the vehicle condition is predicted to enter the pure oil operation condition, and the energy control strategy is to do nothing; When the current vehicle condition is a series operating condition, or the vehicle condition is predicted to be a series operating condition, the energy control strategy is to enter a power following mode, where the power output of the engine follows the power demand of the drive motor, without charging the power battery or discharging the power battery. When the current vehicle condition is a series condition, the vehicle condition is predicted to enter a parallel condition, and the energy control strategy is to do nothing; When the current vehicle condition is a series operating condition, the predicted vehicle condition is to enter a pure electric operating condition, and the energy control strategy is to do nothing.

Images