Energy management method for vehicle

Abstract

The invention relates to the technical field of new energy vehicle detection, in particular to a vehicle energy management method, and the method comprises the steps: obtaining corresponding working condition data of at least one vehicle under different working conditions; segmenting the road of the vehicle in the driving process to obtain road section working condition data corresponding to different road sections; the SOC value, meeting the preset optimal condition, of at least one vehicle under the corresponding working condition is calculated; and training a pre-constructed CNN-LSTM global SOC planning model by using the SOC value and the working condition data meeting a preset optimal condition to obtain a trained CNN-LSTM global SOC planning model, and predicting the SOC value of the vehicle based on the trained CNN-LSTM global SOC planning model to realize energy management of the vehicle. Therefore, the problems that in the related technology, self-adaption to complex traffic environments and working conditions cannot be achieved, the intelligent degree is not high, the energy utilization rate is low, and fuel economy cannot be effectively improved are solved.

Description

Technical Field

[0001] This application relates to the technical field of new energy vehicle detection, and particularly relates to an energy management method for a vehicle. Background Art

[0002] The vehicle combines the advantages of a traditional internal combustion engine and an electric drive system, effectively reducing environmental pollution while providing a long driving range. With the progress of technology, the automotive industry has gradually entered the era of intelligentization and networking, enabling the control strategies of automotive components and systems to be effectively integrated with the ITS (Intelligent Transport System). Integrating real-time vehicle condition information, operating condition data, traffic information, etc. into the energy management strategy of a range-extended hybrid vehicle can significantly improve the operating condition adaptability, thereby improving fuel economy.

[0003] In related technologies, a traffic information-based LSTM (Long Short-Term Memory) vehicle speed prediction model can be constructed based on the intelligent transport system, and the global system-on-chip SOC (State of Charge) can be planned through the target model. Furthermore, the relationship between the adaptive equivalent factor and the system-on-chip SOC can be established, and the optimal control quantity can be obtained by solving the ECMS (Equivalent Consumption Minimization Strategy); alternatively, different types of optimal equivalent factors can be obtained based on the DP algorithm and the ECMS strategy, and then a numerical library of torque, SOC, and equivalent factors under different road section categories and different battery SOCs can be constructed. Furthermore, the optimal equivalent factor adaptable to the operating condition can be obtained by combining the vehicle speed information and the equivalent factor expression, thereby performing energy management for a variable equivalent factor hybrid vehicle.

[0004] However, in related technologies, it is impossible to adapt to complex traffic environments and operating conditions, the degree of intelligentization of vehicle energy scheduling is not high, and the energy utilization rate is low, resulting in the inability to effectively improve fuel economy, which urgently needs to be improved. Summary of the Invention

[0005] This application provides an energy management method for a vehicle to solve the problems in related technologies, such as the inability to adapt to complex traffic environments and operating conditions, low intelligentization, low energy utilization rate, and the inability to effectively improve fuel economy.

[0006] An embodiment of the first aspect of the present application provides an energy management method for a vehicle, which is applied to the off-line training stage of the model. The method includes the following steps: obtaining the working condition data corresponding to at least one vehicle under different working conditions; based on the working condition data, segmenting the road during the driving process of the vehicle to obtain the section working condition data corresponding to different road sections; based on the section working condition data, calculating the SOC value of the at least one vehicle that meets the preset optimal conditions under the corresponding working conditions; using the SOC value that meets the preset optimal conditions and the working condition data to train the pre-constructed CNN (Convolutional Neural Network)-LSTM global SOC planning model to obtain a trained CNN-LSTM global SOC planning model, and predicting the SOC value of the vehicle based on the trained CNN-LSTM global SOC planning model to achieve the energy management of the vehicle.

[0007] Through the above technical solution, the road can be segmented according to the working condition data corresponding to different working conditions obtained by the vehicle, and then the section working condition data corresponding to different road sections can be obtained. Furthermore, the SOC value of the vehicle that meets certain optimal conditions under the corresponding working conditions can be calculated. Then, the pre-constructed CNN-LSTM global SOC planning model can be trained. Thus, the SOC value of the vehicle can be predicted by using the trained CNN-LSTM global SOC planning model to achieve the energy management of the vehicle, enabling the vehicle to dynamically optimize the energy usage strategy according to different working conditions, improving the overall energy efficiency and economy of the range-extended hybrid vehicle. By optimizing the SOC, it is ensured that the actual SOC of the vehicle is as close as possible to the SOC curve that meets certain optimal conditions, optimizing the charging and discharging process of the battery, extending the battery life and reducing energy waste, and reasonably adjusting the power output of the range extender. This can not only optimize the energy management of a single vehicle, but also potentially provide real-time scheduling support for a wider fleet or traffic system in the future, further optimizing the road network and traffic flow, and reducing the overall energy consumption and carbon emissions.

[0008] Optionally, in an embodiment of the present application, segmenting the road during the vehicle's driving based on the operating condition data includes: determining at least one of the average vehicle speed, road segment length, and road segment congestion level during the vehicle's driving based on the operating condition data; judging whether the vehicle speed change rate at different positions is greater than a preset value based on the average vehicle speed and the road segment length, and / or judging whether the road segment congestion level meets a preset condition based on the road segment congestion level and the road segment length; if the vehicle speed change rate is greater than the preset value, segmenting the corresponding road according to the average vehicle speed and the road segment length, and / or if the road segment congestion level meets the preset condition, segmenting the corresponding road according to the road segment congestion level and the road segment length; if the vehicle speed change rate is less than or equal to the preset value, not segmenting the corresponding road, and / or if the road segment congestion level does not meet the preset condition, not segmenting the corresponding road.

[0009] Through the above technical solution, it is possible to judge whether the vehicle speed change rate at different positions is greater than a preset value according to at least one of the average vehicle speed, road segment length, and road segment congestion level during the vehicle's driving. When the vehicle speed change rate is greater than a certain value, segment the corresponding road according to the average vehicle speed and the road segment length, otherwise, do not segment; and / or judge whether the road segment congestion level meets a certain condition. When the road segment congestion level meets a certain condition, segment the corresponding road according to the road segment congestion level and the road segment length, otherwise, do not segment. By calculating the average vehicle speed and congestion level of different road segments, the actual traffic conditions of the road can be more accurately reflected, the data accuracy and refinement can be improved, and then a dynamic adjustment strategy can be formulated to achieve reasonable resource allocation, optimize traffic management and control, and promote the development of intelligent transportation systems.

[0010] Optionally, in an embodiment of the present application, obtaining the operating condition data corresponding to at least one vehicle under different operating conditions includes: obtaining the initial driving data of the at least one vehicle under the corresponding operating condition; obtaining the initial road data of the at least one vehicle under the corresponding operating condition; respectively processing the initial driving data and the initial road data to obtain the corresponding driving data and road data; and obtaining the operating condition data based on the driving data and the road data.

[0011] Through the above technical solution, the initial driving data and initial road data of the vehicle under the corresponding working conditions can be processed, and then the corresponding driving data and road data can be obtained, so as to obtain the working condition data. By performing refined processing on the data, the data dimension can be reduced, the computational complexity can be reduced, while the key information is retained, the model training efficiency is improved, and the model training quality is improved. In addition, the working condition data combines the driving data and the road data, with full coverage of the data dimension, and can better adapt to different driving scenarios, avoiding the overfitting problem caused by single data, and improving the generalization ability in practical applications.

[0012] Optionally, in an embodiment of the present application, before training the pre-constructed CNN-LSTM global SOC planning model using the SOC value and the working condition data that meet the preset optimal conditions, it further includes: constructing an input feature matrix of the CNN-LSTM global SOC planning model using the SOC value and the working condition data that meet the preset optimal conditions; determining the CNN feature extraction layer of the CNN-LSTM global SOC planning model based on the input feature matrix; obtaining an output one-dimensional vector corresponding to the input feature matrix using the CNN feature extraction layer; determining the LSTM network of the CNN-LSTM global SOC planning model based on the output one-dimensional vector; and constructing the CNN-LSTM global SOC planning model using the CNN feature extraction layer and the LSTM network.

[0013] Through the above technical solution, an input feature matrix of the CNN-LSTM global SOC planning model can be constructed using the SOC value and the working condition data that meet certain optimal conditions, and then the CNN feature extraction layer can be determined, and the CNN feature extraction layer can be used to generate an output one-dimensional vector, so as to determine the LSTM network, and then construct the CNN-LSTM global SOC planning model. Through multi-source data fusion, the operating state of the vehicle and the external environment can be more comprehensively reflected, providing richer information for the model. The CNN feature extraction layer automatically extracts the spatial and temporal features in the input feature matrix, captures the change patterns of the SOC value under different working conditions, and improves the expression ability of the model. The LSTM network captures the long-term dependencies in the time series, improves the prediction accuracy, and then combines CNN and LSTM to achieve the global optimal SOC planning.

[0014] Optionally, in an embodiment of the present application, the determining the LSTM network of the CNN-LSTM global SOC planning model based on the output one-dimensional vector includes: determining the input gate of the LSTM network based on the output one-dimensional vector; obtaining the cell state and the hidden state of the LSTM network; determining the forget gate of the LSTM network based on the input gate and the hidden state; determining the output gate of the LSTM network based on the input gate, the cell state, the hidden state, and the forget gate; and obtaining the LSTM network by using the input gate, the cell state, the hidden state, the forget gate, and the output gate.

[0015] Through the above technical solution, the input gate of the LSTM network can be determined based on the output one-dimensional vector, and by obtaining the cell state and the hidden state of the LSTM network, and determining the forget gate and the output gate, the LSTM network can be further constructed. Through the gating mechanism of the LSTM network, the working condition data and the SOC value that meets certain optimal conditions are deeply integrated into the LSTM network design, significantly improving the modeling ability of the LSTM network for the dynamic characteristics of SOC, the adaptability to complex working conditions, and the interpretability, and thus realizing the accurate prediction of SOC and the efficient management of vehicle energy.

[0016] Optionally, in an embodiment of the present application, the training the pre-constructed CNN-LSTM global SOC planning model by using the SOC value that meets the preset optimal condition and the working condition data to obtain the trained CNN-LSTM global SOC planning model includes: obtaining the first-moment estimation exponential decay rate, the second-moment estimation exponential decay rate, and the learning rate of the Adam (Adaptive Moment Estimation) optimizer corresponding to the training of the pre-constructed CNN-LSTM global SOC planning model; and training the pre-constructed CNN-LSTM global SOC planning model based on the first-moment estimation exponential decay rate, the second-moment estimation exponential decay rate, and the learning rate until the pre-constructed CNN-LSTM global SOC planning model meets the preset training condition to obtain the trained CNN-LSTM global SOC planning model.

[0017] Through the above technical solution, the pre-constructed CNN-LSTM global SOC planning model can be trained using the Adam optimizer until certain training conditions are met, and then the trained CNN-LSTM global SOC planning model can be obtained. The Adam optimizer can automatically adjust the learning rate according to different parameters and training stages, avoiding the cumbersome process of manually adjusting the learning rate, improving the training efficiency and model performance. When training the pre-constructed CNN-LSTM global SOC planning model, it can find the optimal solution faster, reduce the training time, effectively handle the problems of gradient disappearance and gradient explosion, and improve the stability and robustness of the model.

[0018] Optionally, in an embodiment of the present application, predicting the SOC value of the vehicle based on the trained CNN-LSTM global SOC planning model to implement the energy management of the vehicle includes: determining the energy consumption per unit distance of the vehicle on the corresponding road section based on the SOC value; using the road section condition data, the SOC value, and the energy consumption per unit distance to construct an energy management database corresponding to the vehicle; and generating an energy management strategy for the vehicle using the energy management database to implement the energy management of the vehicle.

[0019] Through the above technical solution, based on the SOC value, the energy consumption per unit distance of the vehicle on the corresponding road section can be determined, and then an energy management database can be constructed using the road section condition data, the SOC value, and the energy consumption per unit distance, and an energy management strategy for the vehicle can be generated using the energy management database to implement the energy management of the vehicle. The energy management database can store energy consumption information under different road sections and different working conditions, realize the refined monitoring and management of the vehicle's energy consumption, avoid energy waste, improve the scientificity of decision-making, optimize the energy utilization efficiency, and enhance the adaptability and robustness of the strategy.

[0020] An embodiment of the second aspect of the present application provides a method for energy management of a vehicle, which is applied to the model online application stage. The method includes the following steps: obtaining the actual condition data of the vehicle under different working conditions; segmenting the road during the driving process of the vehicle based on the actual condition data to obtain the actual road section condition data corresponding to different road sections; inputting the starting SOC value corresponding to the actual road section condition data into the trained CNN-LSTM global SOC planning model to obtain the ending SOC value of the vehicle on the corresponding road section, where the trained CNN-LSTM global SOC planning model is trained by the starting SOC value and the actual road section condition data; and controlling the power output of the vehicle according to the starting SOC value or the ending SOC value to implement the energy management of the vehicle.

[0021] Through the above technical solution, based on the acquired actual working condition data, the road during the vehicle's driving can be segmented, and then the actual section working condition data corresponding to different road sections can be obtained. The obtained data is used in the trained CNN-LSTM global SOC planning model to output the terminal SOC value of the vehicle on the corresponding road section. Then, the power output of the vehicle is controlled according to the starting SOC value or the terminal SOC value, realizing the energy management of the vehicle. Through segmented prediction, the accumulation of errors is suppressed, the SOC prediction error is reduced, the SOC prediction value is dynamically adjusted according to the actually received working condition data in real time, the prediction delay is reduced, and the real-time requirement is met. Adaptive energy management according to different working conditions can improve driving comfort, extend battery life, and enhance safety.

[0022] The third aspect of the embodiments of the present application provides an energy management device for a vehicle, which is applied to the offline training stage of the model. The device includes: a first acquisition module, configured to acquire the working condition data corresponding to at least one vehicle under different working conditions; a first segmentation module, configured to segment the road during the vehicle's driving based on the working condition data to obtain the section working condition data corresponding to different road sections; a first calculation module, configured to calculate the SOC value that meets the preset optimal condition of the at least one vehicle under the corresponding working condition based on the section working condition data; a prediction module, configured to train a pre-constructed CNN-LSTM global SOC planning model by using the SOC value that meets the preset optimal condition and the working condition data to obtain a trained CNN-LSTM global SOC planning model, and predict the SOC value of the vehicle based on the trained CNN-LSTM global SOC planning model to realize the energy management of the vehicle.

[0023] Through the above technical solution, according to the working condition data corresponding to different working conditions acquired by the vehicle, the road can be segmented, and then the section working condition data corresponding to different road sections can be obtained. Then, the SOC value that meets certain optimal conditions of the vehicle under the corresponding working condition is calculated, and then the pre-constructed CNN-LSTM global SOC planning model is trained. Thus, the trained CNN-LSTM global SOC planning model is used to predict the SOC value of the vehicle, realizing the energy management of the vehicle, enabling the vehicle to dynamically optimize the energy usage strategy according to different working conditions, improving the overall energy efficiency and economy of the range-extended hybrid vehicle, and ensuring that the actual SOC of the vehicle is as close as possible to the SOC curve that meets certain optimal conditions by optimizing the SOC, optimizing the charging and discharging process of the battery, extending the battery life and reducing energy waste, and reasonably adjusting the power output of the range extender. This can not only optimize the energy management of a single vehicle, but also potentially provide real-time scheduling support for a wider fleet or traffic system in the future, further optimizing the road network and traffic flow, and reducing the overall energy consumption and carbon emissions.

[0024] Optionally, in an embodiment of the present application, the first segmentation module includes: a first determination unit configured to determine at least one of an average vehicle speed, a road section length, and a road section congestion level of the vehicle during driving based on the driving condition data; a judgment unit configured to judge whether a vehicle speed change rate at different positions is greater than a preset value based on the average vehicle speed and the road section length, and / or judge whether the road section congestion level meets a preset condition based on the road section congestion level and the road section length; a first segmentation unit configured to segment a corresponding road according to the average vehicle speed and the road section length when the vehicle speed change rate is greater than the preset value, and / or, if the road section congestion level meets the preset condition, segment the corresponding road according to the road section congestion level and the road section length; a second segmentation unit configured to not segment the corresponding road when the vehicle speed change rate is less than or equal to the preset value, and / or, if the road section congestion level does not meet the preset condition, not segment the corresponding road.

[0025] Through the above technical solution, it is possible to judge whether the vehicle speed change rate at different positions is greater than a preset value based on at least one of the average vehicle speed, the road section length, and the road section congestion level of the vehicle during driving, and when the vehicle speed change rate is greater than a certain value, segment the corresponding road according to the average vehicle speed and the road section length, otherwise, do not segment; and / or, judge whether the road section congestion level meets a certain condition, and when the road section congestion level meets a certain condition, segment the corresponding road according to the road section congestion level and the road section length, otherwise, do not segment. By calculating the average vehicle speed and the congestion level of different road sections, the actual traffic conditions of the road can be more accurately reflected, the data accuracy and refinement can be improved, and then a dynamic adjustment strategy can be formulated to achieve reasonable resource allocation, optimize traffic management and control, and promote the development of intelligent transportation systems.

[0026] Optionally, in an embodiment of the present application, the first acquisition module includes: a first acquisition unit configured to acquire initial driving data of the at least one vehicle under corresponding driving conditions; a second acquisition unit configured to acquire initial road data of the at least one vehicle under corresponding driving conditions; a processing unit configured to process the initial driving data and the initial road data respectively to obtain corresponding driving data and road data; a generation unit configured to obtain the driving condition data based on the driving data and the road data.

[0027] Through the above technical solution, the initial driving data and initial road data of the vehicle under the corresponding working conditions can be processed, and then the corresponding driving data and road data can be obtained, so as to obtain the working condition data. By performing refined processing on the data, the data dimension can be reduced, the calculation complexity can be reduced, while the key information is retained, the model training efficiency is improved, and the model training quality is improved. In addition, the working condition data combines the driving data and the road data, and the data dimension is fully covered, which can better adapt to different driving scenarios, avoid the overfitting problem caused by single data, and improve the generalization ability in practical applications.

[0028] Optionally, in an embodiment of the present application, it further includes: a first construction module, configured to construct an input feature matrix of the CNN-LSTM global SOC planning model by using the SOC value and the working condition data that meet the preset optimal conditions before training the pre-constructed CNN-LSTM global SOC planning model by using the SOC value and the working condition data that meet the preset optimal conditions; a first determination module, configured to determine the CNN feature extraction layer of the CNN-LSTM global SOC planning model based on the input feature matrix; a first generation module, configured to obtain an output one-dimensional vector corresponding to the input feature matrix by using the CNN feature extraction layer; a second determination module, configured to determine the LSTM network of the CNN-LSTM global SOC planning model based on the output one-dimensional vector; a second construction module, configured to construct the CNN-LSTM global SOC planning model by using the CNN feature extraction layer and the LSTM network.

[0029] Through the above technical solution, an input feature matrix of the CNN-LSTM global SOC planning model can be constructed by using the SOC value and the working condition data that meet certain optimal conditions, and then the CNN feature extraction layer can be determined, and the output one-dimensional vector can be generated by using the CNN feature extraction layer, so as to determine the LSTM network, and then construct the CNN-LSTM global SOC planning model. By fusing multi-source data, the operating state of the vehicle and the external environment can be more comprehensively reflected, providing richer information for the model. The CNN feature extraction layer automatically extracts the spatial and temporal features in the input feature matrix, captures the change patterns of the SOC value under different working conditions, and improves the expression ability of the model. The LSTM network captures the long-term dependence relationships in the time series, improves the prediction accuracy, and then combines CNN and LSTM to achieve the global optimal SOC planning.

[0030] Optionally, in an embodiment of the present application, the second determination module includes: a second determination unit configured to determine an input gate of the LSTM network based on the output one-dimensional vector; a third acquisition unit configured to acquire a cell state and a hidden state of the LSTM network; a third determination unit configured to determine a forget gate of the LSTM network based on the input gate and the hidden state; a fourth determination unit configured to determine an output gate of the LSTM network based on the input gate, the cell state, the hidden state, and the forget gate; and a second generation unit configured to obtain the LSTM network by using the input gate, the cell state, the hidden state, the forget gate, and the output gate.

[0031] Through the above technical solution, the input gate of the LSTM network can be determined based on the output one-dimensional vector, and by acquiring the cell state and the hidden state of the LSTM network, and determining the forget gate and the output gate, the LSTM network can be further constructed. Through the gating mechanism of the LSTM network, the operating condition data and the SOC value satisfying certain optimal conditions are deeply integrated into the LSTM network design, significantly improving the modeling ability of the LSTM network for the dynamic characteristics of SOC, the adaptability to complex operating conditions, and the interpretability, and thus realizing the accurate prediction of SOC and the efficient management of vehicle energy.

[0032] Optionally, in an embodiment of the present application, the prediction module includes: a fourth acquisition unit configured to acquire an exponential decay rate of the first moment estimate, an exponential decay rate of the second moment estimate, and a learning rate of the Adam optimizer corresponding to training the pre-constructed CNN-LSTM global SOC planning model; and a training unit configured to train the pre-constructed CNN-LSTM global SOC planning model based on the exponential decay rate of the first moment estimate, the exponential decay rate of the second moment estimate, and the learning rate until the pre-constructed CNN-LSTM global SOC planning model meets a preset training condition, so as to obtain the trained CNN-LSTM global SOC planning model.

[0033] Through the above technical solution, the pre-constructed CNN-LSTM global SOC planning model can be trained by using the Adam optimizer until it meets certain training conditions, and then the trained CNN-LSTM global SOC planning model can be obtained. The Adam optimizer can automatically adjust the learning rate according to different parameters and training stages, avoiding the cumbersome process of manually adjusting the learning rate, improving the training efficiency and model performance. When training the pre-constructed CNN-LSTM global SOC planning model, it can find the optimal solution faster, reduce the training time, effectively handle the problems of gradient vanishing and gradient explosion, and improve the stability and robustness of the model.

[0034] Optionally, in an embodiment of the present application, the prediction module includes: a fifth determination unit, configured to determine the energy consumption per unit distance of the vehicle on the corresponding road section based on the SOC value; a third construction unit, configured to construct an energy management database for the vehicle by using the road section condition data, the SOC value, and the energy consumption per unit distance; and a third generation unit, configured to generate an energy management strategy for the vehicle by using the energy management database to implement energy management of the vehicle.

[0035] Through the above technical solution, the energy consumption per unit distance of the vehicle on the corresponding road section can be determined based on the SOC value, and then an energy management database can be constructed by using the road section condition data, the SOC value, and the energy consumption per unit distance, and an energy management strategy for the vehicle can be generated by using the energy management database to implement energy management of the vehicle. The energy management database can store energy consumption information under different road sections and different conditions, realize refined monitoring and management of vehicle energy consumption, avoid energy waste, improve the scientificity of decision-making, optimize energy utilization efficiency, and enhance the adaptability and robustness of the strategy.

[0036] An embodiment of the fourth aspect of the present application provides an energy management device for a vehicle, which is applied to the online application stage of the model. The device includes: a second acquisition module, configured to acquire the actual condition data of the vehicle under different conditions; a second segmentation module, configured to segment the road during the driving process of the vehicle based on the actual condition data to obtain the actual road section condition data corresponding to different road sections; a second generation module, configured to input the starting SOC value corresponding to the actual road section condition data into the trained CNN-LSTM global SOC planning model to obtain the ending SOC value of the vehicle on the corresponding road section, where the trained CNN-LSTM global SOC planning model is trained by the starting SOC value and the actual road section condition data; and a control module, configured to control the power output of the vehicle according to the starting SOC value or the ending SOC value to implement energy management of the vehicle.

[0037] Through the above technical solution, the road during the driving process of the vehicle can be segmented based on the acquired actual condition data, and then the actual road section condition data corresponding to different road sections can be obtained, and the ending SOC value of the vehicle on the corresponding road section can be output by using the trained CNN-LSTM global SOC planning model, and then the power output of the vehicle can be controlled according to the starting SOC value or the ending SOC value to implement energy management of the vehicle. By segmental prediction, error accumulation can be suppressed, the SOC prediction error can be reduced, the SOC prediction value can be dynamically adjusted according to the actually received actual condition data in real time, the prediction delay can be reduced, the real-time requirement can be met, the energy management can be adapted to different conditions, the driving comfort can be improved, the battery life can be extended, and the safety can be enhanced.

[0038] The fifth aspect embodiment of the present application provides a vehicle, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the program to implement the energy management method of the vehicle as described in the above embodiments.

[0039] The sixth aspect embodiment of the present application provides a computer-readable storage medium that stores a computer program, and when the program is executed by a processor, it implements the energy management method of the vehicle as described above.

[0040] The seventh aspect embodiment of the present application provides a computer program product, including a computer program, and when the program is executed, it implements the energy management method of the vehicle as described above.

[0041] The embodiments of the present application can segment the road according to the working condition data corresponding to different working conditions obtained by the vehicle, and then obtain the section working condition data corresponding to different road sections, and then calculate the SOC value that meets certain optimal conditions under the corresponding working conditions of the vehicle, and then train the pre-constructed CNN-LSTM global SOC planning model, so as to use the trained CNN-LSTM global SOC planning model to predict the SOC value of the vehicle, realize the energy management of the vehicle, enable the vehicle to dynamically optimize the energy usage strategy according to different working conditions, improve the overall energy efficiency and economy of the range-extended hybrid vehicle, and by optimizing the SOC, ensure that the actual SOC of the vehicle is as close as possible to the SOC curve that meets certain optimal conditions, optimize the charging and discharging process of the battery, extend the battery life and reduce energy waste, reasonably adjust the power output of the range extender, not only can optimize the energy management of a single vehicle, but also may be able to provide real-time scheduling support for a wider fleet or traffic system in the future, further optimize the road network and traffic flow, and reduce the overall energy consumption and carbon emissions. Thus, the problems in the related art, such as being unable to adapt to complex traffic environments and working conditions, having low intelligence, low energy utilization rate, and being unable to effectively improve fuel economy, are solved.

[0042] The additional aspects and advantages of the present application will be partially given in the following description, partially become obvious from the following description, or be understood through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The above and / or additional aspects and advantages of the present application will become obvious and easy to understand from the following description of the embodiments in conjunction with the drawings, where: Figure 1 is a flowchart of an energy management method for a vehicle provided according to an embodiment of the present application; Figure 2 is a flowchart of obtaining working condition data provided according to an embodiment of the present application; Figure 3Schematic diagram of a CNN-LSTM network provided according to an embodiment of the present application; Figure 4 Schematic diagram of a structure for training a pre-trained CNN-LSTM global SOC planning model provided according to an embodiment of the present application; Figure 5 Block diagram of an energy management device for a vehicle provided according to an embodiment of the present application; Figure 6 Flowchart of an energy management method for a vehicle provided according to another embodiment of the present application; Figure 7 Flowchart of the working principle of an energy management method for a vehicle provided according to another embodiment of the present application; Figure 8 Block diagram of an energy management device for a vehicle provided according to another embodiment of the present application; Figure 9 Schematic diagram of a vehicle provided according to an embodiment of the present application.

[0044] Reference numerals: Among them, 50 - energy management device of the vehicle; 501 - first acquisition module, 502 - first segmentation module, 503 - first calculation module, 504 - prediction module; 80 - energy management device of the vehicle; 801 - second acquisition module, 802 - second segmentation module, 803 - second generation module, 804 - control module; 901 - memory, 902 - processor, 903 - communication interface. Detailed implementation manners

[0045] The embodiments of the present application will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals indicate the same or similar elements or elements with the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present application, and should not be construed as a limitation to the present application.

[0046] The energy management method of the vehicle according to the embodiments of the present application will be described below with reference to the accompanying drawings. In view of the problems mentioned in the above background art, such as the inability to adapt to complex traffic environments and working conditions, low intelligence level, low energy utilization rate, and ineffective improvement of fuel economy, the present application provides an energy management method for a vehicle. In this method, according to the working condition data corresponding to different working conditions obtained by the vehicle, the road can be segmented, and then the section working condition data corresponding to different road sections can be obtained. Furthermore, the SOC value that meets certain optimal conditions under the corresponding working conditions can be calculated, and then the pre-constructed CNN-LSTM global SOC planning model can be trained. Thus, the trained CNN-LSTM global SOC planning model can be used to predict the SOC value of the vehicle, realizing the energy management of the vehicle, enabling the vehicle to dynamically optimize the energy usage strategy according to different working conditions, improving the overall energy efficiency and economy of the range-extended hybrid vehicle, and by optimizing the SOC, ensuring that the actual SOC of the vehicle is as close as possible to the SOC curve that meets certain optimal conditions, optimizing the charging and discharging process of the battery, extending the battery life and reducing energy waste, and reasonably adjusting the power output of the range extender. This can not only optimize the energy management of a single vehicle, but also potentially provide real-time scheduling support for a wider fleet or traffic system in the future, further optimizing the road network and traffic flow, and reducing the overall energy consumption and carbon emissions. Thereby, the problems in the related art, such as the inability to adapt to complex traffic environments and working conditions, low intelligence level, low energy utilization rate, and ineffective improvement of fuel economy, are solved.

[0047] Specifically, Figure 1 FIG. is a flowchart of an energy management method for a vehicle according to an embodiment of the present application.

[0048] As Figure 1 shown, this energy management method for a vehicle is applied to the model offline training stage, and the method includes the following steps: In step S101, obtain the working condition data corresponding to different working conditions of at least one vehicle.

[0049] It can be understood that in the embodiments of the present application, the working condition data can include, but is not limited to, the driving data for collecting the driver's travel working conditions using a GPS (Global Positioning System) receiver, the road data provided by a map API (Application Programming Interface), etc. The present application does not make specific limitations.

[0050] Furthermore, in the embodiments of the present application, different working conditions may include, but are not limited to, urban working conditions, highway working conditions, mountainous working conditions, etc., and the present application does not make specific limitations. Among them, in urban working conditions, the road is relatively congested, traffic lights are relatively dense, the vehicle travels at a low speed, and starts and stops frequently. In highway working conditions, the road is relatively flat, traffic lights are relatively sparse, and the vehicle travels steadily at a high speed. In mountainous working conditions, there are a large number of uphill and downhill roads, and the vehicle ensures safe driving.

[0051] As a possible implementation manner, the embodiments of the present application may obtain the working condition data corresponding to at least one vehicle under different working conditions.

[0052] Exemplarily, the embodiments of the present application may obtain the working condition data through the driving data collected by the GPS receiver and the road data provided by the map API.

[0053] Optionally, in an embodiment of the present application, obtaining the working condition data corresponding to at least one vehicle under different working conditions includes: obtaining the initial driving data of at least one vehicle under the corresponding working condition; obtaining the initial road data of at least one vehicle under the corresponding working condition; respectively processing the initial driving data and the initial road data to obtain the corresponding driving data and road data; and obtaining the working condition data based on the driving data and the road data.

[0054] It can be understood that in the embodiments of the present application, the initial driving data may include, but is not limited to, the vehicle speed, acceleration, battery SOC, rotation speed, etc., and the present application does not make specific limitations. The initial road data may include, but is not limited to, congestion conditions, road length, remaining driving mileage, signal light positions, etc., and the present application does not make specific limitations.

[0055] In the actual execution process, the embodiments of the present application may use the GPS receiver to collect the initial driving data, process it, and extract the working condition curves, such as the change trends of vehicle speed, acceleration, battery SOC, etc. over time or mileage, so as to obtain the corresponding driving data. Use the map API to obtain the initial road data, where the initial road data may include, but is not limited to, road congestion conditions (unobstructed, wake-up, congested, severely congested), road length, etc., and the present application does not make specific limitations, and process it to obtain the corresponding road data. Further, the embodiments of the present application generate the working condition data according to the driving data and the road data.

[0056] Exemplarily, in combination with Figure 2As shown in the figure, the embodiments of the present application can collect initial driving data by using a GPS receiver, including the average vehicle speed, battery SOC, etc., which are not specifically limited in the present application. At the same time, initial road data provided by the map API is collected, including the congestion situation of road sections (unobstructed, slow-moving, congested, severely congested), road section length, etc., which are not specifically limited in the present application. Further, the embodiments of the present application process the initial driving data and the initial road data, and combine historical data to construct working condition data including 30 groups of working conditions. Among them, the working condition data can cover most of the daily use scenarios.

[0057] In step S102, based on the working condition data, the road during the vehicle's driving is segmented to obtain the road section working condition data corresponding to different road sections.

[0058] It can be understood that the embodiments of the present application can segment the road according to the vehicle speed change rate, road section length, and road section congestion level. Specifically, those skilled in the art can set it according to the actual situation, and the present application does not make specific limitations.

[0059] In the actual execution process, the embodiments of the present application can segment the road during the vehicle's driving, and then obtain the road section working condition data corresponding to different road sections.

[0060] Optionally, in an embodiment of the present application, segmenting the road during the vehicle's driving based on the working condition data includes: determining at least one of the average vehicle speed, road section length, and road section congestion level during the vehicle's driving based on the working condition data; judging whether the vehicle speed change rate at different positions is greater than a preset value based on the average vehicle speed and road section length, and / or judging whether the road section congestion level meets a preset condition based on the road section congestion level and road section length; if the vehicle speed change rate is greater than the preset value, segment the corresponding road according to the average vehicle speed and road section length, and / or if the road section congestion level meets the preset condition, segment the corresponding road according to the road section congestion level and road section length; if the vehicle speed change rate is less than or equal to the preset value, the corresponding road is not segmented, and / or if the road section congestion level does not meet the preset condition, the corresponding road is not segmented.

[0061] In some embodiments, the embodiments of the present application can segment the road according to the average vehicle speed, road section length, and road section congestion level during the vehicle's driving, and the content can be: In some embodiments, the embodiments of the present application can determine whether the vehicle speed change rate at different positions is greater than a certain value based on the average vehicle speed and the road section length, and when the vehicle speed change rate is greater than a certain value, segment the corresponding road according to the average vehicle speed and the road section length; when the vehicle speed change rate is less than or equal to the preset value, the corresponding road is not segmented. Among them, the certain value can be set by those skilled in the art according to the actual situation, and the present application does not make specific limitations.

[0062] Exemplarily, the embodiments of the present application can improve the method for segmenting road sections divided by the static API map, introduce a dynamic adaptive segmentation mechanism, and then segment the road, and its content is as follows: Among them, the embodiments of the present application can extract road section features based on the working condition data, and the information included in each API road section can but is not limited to: , Among them, represents the average vehicle speed of the road section, represents the road section length of the road section, represents the road section congestion level of the road section.

[0063] Furthermore, the embodiments of the present application can segment the road during the vehicle's driving process into multiple sub-sections, and the judgment conditions are as follows: If the vehicle speed change rate between two adjacent sampling points is greater than a certain value, such as 0.2, then segmentation is performed, and its expression can but is not limited to be expressed as: , Among them, represents the average vehicle speed of the previous sampling point, represents the average vehicle speed of the current sampling point.

[0064] The input features of each adaptive road section are reconstructed as: , and then used to input into the pre-constructed CNN-LSTM global SOC planning model to predict the SOC value.

[0065] In some embodiments, the embodiments of the present application can determine whether the road section congestion level meets certain conditions based on the road section congestion level and the road section length. When the road section congestion level meets certain conditions, segment the corresponding road according to the road section congestion level and the road section length; when the road section congestion level does not meet certain conditions, the corresponding road is not segmented. Among them, the certain conditions can be set by those skilled in the art according to the actual situation, and the present application does not make specific limitations.

[0066] Exemplarily, embodiments of the present application can improve the method for segmenting road sections of a static API map, introduce a dynamic adaptive segmentation mechanism, and further segment the road. The content is as follows: Among them, embodiments of the present application can extract road section features based on operating condition data. The information included in each API road section can include but is not limited to: , Among them, represents the average vehicle speed of the road section, represents the road section length of the road section, represents the road section congestion level of the road section.

[0067] Furthermore, embodiments of the present application can segment the road during the vehicle's driving process into multiple sub-sections. The judgment conditions are as follows: If the road section congestion level meets certain conditions, such as .

[0068] The input features of each adaptive road section are reconstructed as: , and then used to input into a pre-constructed CNN-LSTM global SOC planning model to predict the SOC value.

[0069] In step S103, based on the road section operating condition data, calculate the SOC value of at least one vehicle that meets the preset optimal conditions under the corresponding operating conditions.

[0070] It can be understood that in embodiments of the present application, a certain optimal condition can be understood as the SOC value corresponding to the minimum vehicle energy consumption, which can be specifically set by those skilled in the art according to the actual situation, and the present application does not make specific limitations.

[0071] As a possible implementation manner, embodiments of the present application can calculate the SOC value that meets certain optimal conditions under the corresponding operating conditions according to the operating condition data obtained by the vehicle using the PMP (Pontryagin's Minimum Principle) algorithm, and use the SOC value that meets certain optimal conditions as the training sample for training the pre-constructed CNN-LSTM global SOC planning model.

[0072] Optionally, in an embodiment of the present application, before training a pre-constructed CNN-LSTM global SOC planning model using the SOC value and operating condition data that meet the preset optimal conditions, it further includes: constructing an input feature matrix of the CNN-LSTM global SOC planning model using the SOC value and operating condition data that meet the preset optimal conditions; determining the CNN feature extraction layer of the CNN-LSTM global SOC planning model based on the input feature matrix; obtaining an output one-dimensional vector corresponding to the input feature matrix using the CNN feature extraction layer; determining the LSTM network of the CNN-LSTM global SOC planning model based on the output one-dimensional vector; and constructing the CNN-LSTM global SOC planning model using the CNN feature extraction layer and the LSTM network.

[0073] It can be understood that in the embodiments of the present application, the CNN-LSTM global SOC planning model may but is not limited to include a CNN feature extraction layer, an LSTM network, a fully connected layer, etc., and the present application does not make specific limitations. Among them, in the embodiments of the present application, a Dropout layer is set between the output of the LSTM network and the fully connected layer, and then a regularization operation is introduced, so as to effectively avoid the LSTM network learning too fast and overfitting the training set, which is particularly important especially when the samples are limited. For example, in the CNN-LSTM global SOC planning model of the embodiments of the present application, Dropout(0.3) is introduced between the output of the LSTM network and the fully connected layer to suppress the overfitting risk, and the position is as follows: LSTM(64) → Dropout(0.3) → fully connected layer(1). Among them, the Dropout layer is not specific to a certain network (such as the LSTM network or the CNN-LSTM global SOC planning model), but a general regularization method that can be inserted between any two layers.

[0074] In some embodiments, the process of constructing the CNN-LSTM global SOC planning model in the embodiments of the present application is as follows: First, the embodiments of the present application construct an input feature matrix of the CNN-LSTM global SOC planning model using the SOC value and operating condition data that meet certain optimal conditions, then determine the CNN feature extraction layer, thereby outputting a one-dimensional vector, and use the output one-dimensional vector to determine the LSTM network, and then construct the CNN-LSTM global SOC planning model.

[0075] Exemplarily, in the embodiments of the present application, for each road section, an input time series with a length of and a time step of can be constructed for the input feature matrix, and its expression may but is not limited to: , where, represents the average vehicle speed of the current section, represents the length of the current section, Indicates the starting SOC, Indicates the remaining driving range, Indicates the congestion level, Indicates the global starting SOC.

[0076] Furthermore, the embodiment of the present application uses a CNN feature extraction layer to determine the output one-dimensional vector corresponding to the input feature matrix. It can be understood that the embodiment of the present application can use the CNN feature extraction layer to convert the input feature matrix into a two-dimensional tensor with a size of Extract local patterns through one-dimensional convolution operation, and its expression can be but is not limited to: , Among them, Indicates the input, Indicates the convolution kernel, Indicates the filter.

[0077] Furthermore, the output dimension of the local pattern is , and then flattened into an output one-dimensional vector , and the output one-dimensional vector is input into the LSTM network for time dimension modeling, and then the final time series encoding vector is obtained. Among them, the expression of the output of the LSTM network can be but is not limited to: , Furthermore, the embodiment of the present application can predict the SOC value at the end of the next section through a fully connected layer regression, and its expression can be but is not limited to: .

[0078] Optionally, in an embodiment of the present application, based on the output one-dimensional vector, determine the input gate of the LSTM network; obtain the cell state and hidden state of the LSTM network; based on the input gate and hidden state, determine the forget gate of the LSTM network; based on the input gate, cell state, hidden state and forget gate, determine the output gate of the LSTM network; use the input gate, cell state, hidden state, forget gate and output gate to obtain the LSTM network.

[0079] It should be noted that the embodiment of the present application can introduce the structure of the LSTM network in combination with Figure 3 shown. Among them, the LSTM network can remember information with a long time interval when processing sequence data, and can remember historical information and apply it to the calculation of the current output.

[0080] Furthermore, in the embodiment of the present application, in Figure 3 , , Are respectively At the moment and The next moment The input of the moment LSTM network, , , are divided into the moment, the moment, and the cell state of the moment LSTM network, , , are respectively the moment, the moment, and the output of the moment LSTM network, , are respectively the moment and the state of controlling data transfer in the moment LSTM network. These four variables respectively become the input gate, cell state, output gate, and forget gate of the LSTM neural network, is function.

[0081] Among them, the formula for the working mechanism of the input gate of the LSTM network in the embodiments of this application is as follows, including the current sequence and the previous moment sequence , and its expression can be but is not limited to: , The expression of the working mechanism of the forget gate can be but is not limited to: , The expression of the working mechanism of the output gate can be but is not limited to: , Among them, represents the input of the moment LSTM network, represents the output of the moment LSTM network, represents function, , , , respectively represent different weight coefficients of the LSTM network, , , , respectively represent the corresponding biases of the LSTM network, represents the cell state of the moment LSTM network.

[0082] Further, in the embodiments of the present application, the LSTM global SOC planning model can predict and optimize the SOC reference trajectory. Combining the structure and working mechanism of the LSTM network, the input, cell state, output, and forget gate are introduced.

[0083] Input variable represents the input at time, which may but is not limited to including road congestion , average speed of the road section , length of the road section , SOC value at the start of the road section , total remaining driving mileage and starting SOC value of the working condition ; The input variable is similar to and represents the input at time, corresponding to the feature vector of the next road section.

[0084] Cell state represents the cell state at time, which is the state of the memory unit in the LSTM network and is used to store historical information; represents the cell state at time, which is the cell state updated by the current input and the previous state; represents the cell state at time, which is the cell state updated by the next input.

[0085] Hidden state represents the hidden state at time, which is the output of the LSTM network and is used to be passed to the next time step; represents at the hidden state at time, which is the output state of the current time step and reflects the comprehensive result of the current input and historical information; represents the hidden state at time, which is the output state of the next time step.

[0086] Forget gate represents at the forget gate at time, which is used to determine which information in the cell state needs to be forgotten; represents the forget gate at time, which is used to determine which information needs to be forgotten in the next time step.

[0087] Optionally, in an embodiment of the present application, the pre-constructed CNN-LSTM global SOC planning model is trained using the SOC value and operating condition data that meet the preset optimal conditions to obtain a trained CNN-LSTM global SOC planning model, including: obtaining the first-moment estimation exponential decay rate, second-moment estimation exponential decay rate, and learning rate of the Adam optimizer corresponding to the pre-constructed CNN-LSTM global SOC planning model; based on the first-moment estimation exponential decay rate, second-moment estimation exponential decay rate, and learning rate, training the pre-constructed CNN-LSTM global SOC planning model until the pre-constructed CNN-LSTM global SOC planning model meets the preset training conditions to obtain a trained CNN-LSTM global SOC planning model.

[0088] It can be understood that in the embodiment of the present application, the Adam optimizer can optimize all trainable parameters in the CNN-LSTM global SOC planning model, including the convolution kernel weights and biases of the convolutional layer, the weight matrices and bias terms of each gating unit in the LSTM network structure, and the output weights and biases of the fully connected layer. By gradually minimizing the loss function, the fitting ability and convergence efficiency of the model in the SOC prediction task are effectively improved.

[0089] Among them, the Adam optimizer used in the embodiment of the present application combines the advantages of the momentum method and RMSProp, can achieve adaptive gradient adjustment, and the set core parameters can include but are not limited to the first-moment estimation exponential decay rate, second-moment estimation exponential decay rate, and learning rate, etc., and the present application does not make specific limitations.

[0090] Among them, the first-moment estimation exponential decay rate is used to control the momentum of the gradient, and it can be ; the second-moment estimation exponential decay rate reflects the weighted average of the gradient square, and it can be ; a small constant to prevent the denominator from being zero, for numerical stability; the initial learning rate is set to , and an exponential decay strategy is adopted to dynamically adjust the learning rate: , where , is a number. Among them, in the embodiment of the present application, a larger learning rate is used in the initial stage of training to accelerate convergence, and the step size is reduced in the later stage to refine the weights.

[0091] It should be noted that in the embodiment of the present application, a Dropout layer is set between the output of the LSTM network and the fully connected layer, and the Dropout ratio is 0.3. During the training phase, 30% of the neuron outputs are randomly discarded to enhance the robustness of the model. After each round of training, the loss of the validation set is monitored. If the validation error does not decrease for 10 consecutive rounds, the training process is terminated in advance to avoid overfitting.

[0092] In addition, in the embodiment of the present application, to prevent the problem of gradient explosion, truncation is performed when the norm of the gradient during backpropagation is greater than 5.0. Its expression can be but is not limited to: , Thereby ensuring the stability of model parameter update and numerical convergence.

[0093] Through the above optimization strategy, the model training time is shortened by about 15%, and the mean square error loss of SOC prediction on the validation set is reduced by about 7.2% on average.

[0094] Among them, the expression of the mean square error loss function can be but is not limited to: , Among them, represents the true value of SOC, represents the predicted value of SOC.

[0095] Furthermore, the embodiment of the present application uses the Adam optimizer for training, with an initial learning rate of 0.001, and a built-in momentum mechanism to improve the convergence speed.

[0096] In the validation set of the embodiment of the present application, the average error of SOC prediction drops by 3.8%. Especially in complex working conditions, it is more sensitive to the trend change of SOC, improving the prediction accuracy and robustness of the model.

[0097] In summary, the embodiment of the present application can use the Adam optimizer to train the pre-constructed CNN-LSTM global SOC planning model until the pre-constructed CNN-LSTM global SOC planning model meets certain training conditions, and then obtain the trained CNN-LSTM global SOC planning model. Among them, the certain training conditions can be set by those skilled in the art according to the actual situation, and the present application does not make specific limitations.

[0098] In step S104, the pre-constructed CNN-LSTM global SOC planning model is trained using the SOC values and working condition data that meet the preset optimal conditions to obtain the trained CNN-LSTM global SOC planning model, and the SOC value of the vehicle is predicted based on the trained CNN-LSTM global SOC planning model to achieve the energy management of the vehicle.

[0099] As a possible implementation manner, embodiments of the present application can train a pre-constructed CNN-LSTM global SOC planning model based on working condition data using SOC values that meet preset optimal conditions, and then obtain a trained CNN-LSTM global SOC planning model, and then predict the SOC value of the vehicle, so as to achieve energy management of the vehicle.

[0100] Exemplarily, embodiments of the present application combine Figure 4 as shown, to predict the SOC value, and the main content is as follows: Step S401: Construct training samples.

[0101] Among them, embodiments of the present application can set the initial SOC values to be respectively based on 30 groups of working condition data, with the termination SOC being 0.3, use the PMP algorithm to solve the SOC values that meet certain optimal conditions under different starting SOCs for each group of working conditions, and use the obtained 510 groups of SOC values that meet certain optimal conditions as the training samples of the pre-constructed CNN-LSTM global SOC planning model.

[0102] Step S402: Set model parameters.

[0103] Among them, embodiments of the present application segment the divided road sections according to the map API for each group of SOC values that meet certain optimal conditions (each working condition segment is numbered ), and adopt a variable step size method to predict the termination SOC at the end of each road section. Set the inputs of the model to be , , , , and , and set the output of the model to be . Set the learning rate of the model to be 0.001, and the step size to be the time interval of two road sections. Specifically, those skilled in the art can set it according to the actual situation, and the present application does not make specific limitations.

[0104] Step S403: Train the model.

[0105] Among them, embodiments of the present application can train the pre-constructed CNN-LSTM global SOC planning model based on the training samples and model parameters, update the weight gradients of the pre-constructed CNN-LSTM global SOC planning model, and obtain a trained CNN-LSTM global SOC planning model.

[0106] Step S404: Predict the SOC value.

[0107] Among them, the embodiments of the present application can use the trained CNN-LSTM global SOC planning model to predict the end SOC at the end of each road section and perform interpolation, that is, obtain the SOC curve of the global continuous time. .

[0108] Optionally, in an embodiment of the present application, the SOC value of the vehicle is predicted based on the trained CNN-LSTM global SOC planning model to implement the energy management of the vehicle, including: determining the energy consumption per unit distance of the vehicle on the corresponding road section based on the SOC value; constructing an energy management database corresponding to the vehicle by using the road section condition data, the SOC value and the energy consumption per unit distance; and generating an energy management strategy for the vehicle by using the energy management database to implement the energy management of the vehicle.

[0109] It can be understood that the embodiments of the present application construct an energy management database to store the ternary data of "condition data - SOC - energy consumption" to support the model training and generalization ability expansion. Among them, the sample structure of the ternary data can be but is not limited to: , Among them, represents the input feature vector, constructed as described above; represents the optimal SOC value of the corresponding segment; represents the energy consumption per unit distance on the corresponding road section, with the unit of Wh / km); represents the total number of road sections.

[0110] In addition, the energy management database can include but is not limited to 60 sets of urban road samples, 50 sets of highway samples, and 40 sets of suburban / rural samples, with a total of 150 sets of total samples, a total mileage of more than 1000 km, and data with a sampling period of 5 seconds. The present application does not make specific limitations. Furthermore, the embodiments of the present application can upload the working condition trajectory, the actual SOC curve and the energy consumption to the server after each complete user path operation, automatically match the labels and file them into the energy management database, and then generate an energy management strategy for the vehicle by using the energy management database, so as to implement the energy management of the vehicle.

[0111] According to the energy management method of a vehicle proposed in an embodiment of the present application, the road can be segmented based on the working condition data corresponding to different working conditions obtained by the vehicle, so as to obtain the section working condition data corresponding to different road sections. Then, the SOC value of the vehicle that meets certain optimal conditions under the corresponding working conditions can be calculated. Furthermore, the pre-constructed CNN-LSTM global SOC planning model can be trained, so as to predict the SOC value of the vehicle by using the trained CNN-LSTM global SOC planning model, realizing the energy management of the vehicle, enabling the vehicle to dynamically optimize the energy usage strategy according to different working conditions, improving the overall energy efficiency and economy of the range-extended hybrid vehicle, and by optimizing the SOC, ensuring that the actual SOC of the vehicle is as close as possible to the SOC curve that meets certain optimal conditions, optimizing the charging and discharging process of the battery, extending the battery life and reducing energy waste, and reasonably adjusting the power output of the range extender. This can not only optimize the energy management of a single vehicle, but also potentially provide real-time scheduling support for a wider fleet or traffic system in the future, further optimizing the road network and traffic flow, and reducing the overall energy consumption and carbon emissions. Thus, the problems in the related technology, such as being unable to adapt to complex traffic environments and working conditions, having low intelligence, low energy utilization rate, and being unable to effectively improve fuel economy, are solved.

[0112] Next, a vehicle energy management device proposed according to an embodiment of the present application will be described with reference to the accompanying drawings.

[0113] Figure 5 It is a block diagram of a vehicle energy management device provided according to an embodiment of the present application.

[0114] As Figure 5 shown, the vehicle energy management device 50 is applied to the model offline training stage. Among them, the device 50 includes: a first acquisition module 501, a first segmentation module 502, a first calculation module 503, and a prediction module 504.

[0115] Among them, the first acquisition module 501 is used to acquire the working condition data corresponding to at least one vehicle under different working conditions.

[0116] The first segmentation module 502 is used to segment the road during the vehicle's driving process based on the working condition data, so as to obtain the section working condition data corresponding to different road sections.

[0117] The first calculation module 503 is used to calculate the SOC value of at least one vehicle that meets the preset optimal conditions under the corresponding working conditions based on the section working condition data.

[0118] The prediction module 504 is configured to train a pre-constructed CNN-LSTM global SOC planning model by using the SOC value and driving condition data that meet the preset optimal conditions, so as to obtain a trained CNN-LSTM global SOC planning model, and predict the SOC value of the vehicle based on the trained CNN-LSTM global SOC planning model, so as to implement the energy management of the vehicle.

[0119] Optionally, in an embodiment of the present application, the first segmentation module 502 includes: a first determination unit, a judgment unit, a first segmentation unit, and a second segmentation unit.

[0120] Wherein, the first determination unit is configured to determine at least one of the average vehicle speed, road section length, and road section congestion level of the vehicle during driving based on the driving condition data.

[0121] The judgment unit is configured to judge whether the vehicle speed change rate at different positions is greater than a preset value based on the average vehicle speed and the road section length, and / or judge whether the road section congestion level meets the preset conditions based on the road section congestion level and the road section length.

[0122] The first segmentation unit is configured to segment the corresponding road according to the average vehicle speed and the road section length when the vehicle speed change rate is greater than the preset value, and / or segment the corresponding road according to the road section congestion level and the road section length if the road section congestion level meets the preset conditions.

[0123] The second segmentation unit is configured not to segment the corresponding road when the vehicle speed change rate is less than or equal to the preset value, and / or not to segment the corresponding road if the road section congestion level does not meet the preset conditions.

[0124] Optionally, in an embodiment of the present application, the first acquisition module 501 includes: a first acquisition unit, a second acquisition unit, a processing unit, and a generation unit.

[0125] Wherein, the first acquisition unit is configured to acquire the initial driving data of at least one vehicle under the corresponding driving conditions.

[0126] The second acquisition unit is configured to acquire the initial road data of at least one vehicle under the corresponding driving conditions.

[0127] The processing unit is configured to process the initial driving data and the initial road data respectively to obtain the corresponding driving data and road data.

[0128] The generation unit is configured to obtain the driving condition data based on the driving data and the road data.

[0129] Optionally, in an embodiment of the present application, it further includes: a first construction module, a first determination module, a first generation module, a second determination module, and a second construction module.

[0130] Among them, the first construction module is used to construct the input feature matrix of the CNN-LSTM global SOC planning model by using the SOC value and working condition data that meet the preset optimal conditions before training the pre-constructed CNN-LSTM global SOC planning model with the SOC value and working condition data that meet the preset optimal conditions.

[0131] The first determination module is used to determine the CNN feature extraction layer of the CNN-LSTM global SOC planning model based on the input feature matrix.

[0132] The first generation module is used to obtain the output one-dimensional vector corresponding to the input feature matrix by using the CNN feature extraction layer.

[0133] The second determination module is used to determine the LSTM network of the CNN-LSTM global SOC planning model based on the output one-dimensional vector.

[0134] The second construction module is used to construct the CNN-LSTM global SOC planning model by using the CNN feature extraction layer and the LSTM network.

[0135] Optionally, in an embodiment of the present application, the second determination module includes: a second determination unit, a third acquisition unit, a third determination unit, a fourth determination unit, and a second generation unit.

[0136] Among them, the second determination unit is used to determine the input gate of the LSTM network based on the output one-dimensional vector.

[0137] The third acquisition unit is used to acquire the cell state and hidden state of the LSTM network.

[0138] The third determination unit is used to determine the forget gate of the LSTM network based on the input gate and the hidden state.

[0139] The fourth determination unit is used to determine the output gate of the LSTM network based on the input gate, the cell state, the hidden state, and the forget gate.

[0140] The second generation unit is used to obtain the LSTM network by using the input gate, the cell state, the hidden state, the forget gate, and the output gate.

[0141] Optionally, in an embodiment of the present application, the prediction module 504 includes: a fourth acquisition unit and a training unit.

[0142] Among them, the fourth acquisition unit is used to acquire the first-moment estimation exponential decay rate, the second-moment estimation exponential decay rate, and the learning rate of the Adam optimizer corresponding to training the pre-constructed CNN-LSTM global SOC planning model.

[0143] A training unit is configured to train a pre-constructed CNN-LSTM global SOC planning model based on a first-order moment estimated exponential decay rate, a second-order moment estimated exponential decay rate, and a learning rate until the pre-constructed CNN-LSTM global SOC planning model meets a preset training condition, so as to obtain a trained CNN-LSTM global SOC planning model.

[0144] Optionally, in an embodiment of the present application, the prediction module 504 includes: a fifth determination unit, a third construction unit, and a third generation unit.

[0145] Among them, the fifth determination unit is configured to determine the energy consumption per unit distance of the vehicle on the corresponding road section based on the SOC value.

[0146] The third construction unit is configured to construct an energy management database corresponding to the vehicle by using the road section working condition data, the SOC value, and the energy consumption per unit distance.

[0147] The third generation unit is configured to generate an energy management strategy for the vehicle by using the energy management database to implement the energy management of the vehicle.

[0148] It should be noted that the foregoing explanation of the embodiments of the energy management method for vehicles is also applicable to the energy management device for vehicles in this embodiment, and will not be elaborated here.

[0149] According to the energy management device for vehicles proposed in the embodiments of the present application, the road can be segmented according to the working condition data corresponding to different working conditions obtained by the vehicle, and then the road section working condition data corresponding to different road sections can be obtained. Furthermore, the SOC value of the vehicle that meets certain optimal conditions under the corresponding working conditions can be calculated, and then the pre-constructed CNN-LSTM global SOC planning model can be trained. Thus, the SOC value of the vehicle can be predicted by using the trained CNN-LSTM global SOC planning model to implement the energy management of the vehicle, so that the vehicle can dynamically optimize the energy usage strategy according to different working conditions, improve the overall energy efficiency and economy of the range-extended hybrid vehicle, and by optimizing the SOC, ensure that the actual SOC of the vehicle is as close as possible to the SOC curve that meets certain optimal conditions, optimize the charging and discharging process of the battery, extend the service life of the battery and reduce energy waste, and reasonably adjust the power output of the range extender. This can not only optimize the energy management of a single vehicle, but also potentially provide real-time scheduling support for a wider fleet or transportation system in the future, further optimize the road network and traffic flow, and reduce overall energy consumption and carbon emissions. Thereby, the problems in the related art, such as being unable to adapt to complex traffic environments and working conditions, having low intelligence, low energy utilization rate, and being unable to effectively improve fuel economy, are solved.

[0150] The above embodiments describe the offline training stage of the model. Next, the embodiments of the online application stage of the model will be described.

[0151] Figure 6 A flowchart of an energy management method for a vehicle provided according to another embodiment of the present application.

[0152] As Figure 6 shown, the energy management method of the vehicle is applied to the online application stage of the model. Among them, the method includes the following steps: In step S601, actual condition data of the vehicle under different working conditions is obtained.

[0153] In some embodiments, the embodiments of the present application can collect the actual condition data of the vehicle under different working conditions according to the API map information.

[0154] Exemplarily, in the embodiments of the present application, the starting position of vehicle A is B, the ending position is B', and its working condition is an urban working condition. Furthermore, the embodiments of the present application can use the API map to obtain the actual condition data of the vehicle in real time.

[0155] In step S602, based on the actual condition data, the road during the driving process of the vehicle is segmented to obtain actual section condition data corresponding to different road sections.

[0156] In some embodiments, the embodiments of the present application can segment the road based on the actual condition data, and then obtain the actual section condition data corresponding to different road sections.

[0157] Exemplarily, in the embodiments of the present application, vehicle A can segment the road BB', divide it into three road sections, and then according to the actual section condition data corresponding to different road sections, where section 1 is unobstructed, section 2 is slow moving, and section 3 is congested, and the corresponding actual section condition data is also different.

[0158] In step S603, the starting SOC value corresponding to the actual section condition data is input into the trained CNN-LSTM global SOC planning model to obtain the ending SOC value of the vehicle on the corresponding road section, where the trained CNN-LSTM global SOC planning model is trained by the starting SOC value and the actual section condition data.

[0159] In some embodiments, the embodiments of the present application can extract the congestion situation of each road section , average speed , section length , starting SOC value at the start of the section , total remaining driving mileage and starting SOC value of the working condition As the input of the trained CNN-LSTM global SOC planning model, the starting SOC value of the next road section is then obtained by using the trained CNN-LSTM global SOC planning. .

[0160] In step S604, the power output of the vehicle is controlled according to the starting SOC value or the ending SOC value to achieve the energy management of the vehicle.

[0161] In some embodiments, the embodiments of the present application can calculate the ending SOC at the end of each road section and perform interpolation to obtain the global continuous-time SOC curve. , adjust the power output of the range extender, control the actual SOC to follow the SOC reference curve, achieve the optimal allocation of energy, and achieve the energy management of the vehicle.

[0162] Next, in combination with a specific embodiment, the working principle of the vehicle energy management method proposed by the embodiments of the present application is introduced.

[0163] Among them, Figure 7 is a flowchart of the working principle of the vehicle energy management method provided according to another embodiment of the present application.

[0164] Offline training stage: Step S701: Obtain driving condition data.

[0165] Among them, the embodiments of the present application can combine Figure 2 the method shown to obtain driving condition data.

[0166] Step S702: Based on the driving condition data, calculate the SOC value that meets certain optimal conditions.

[0167] Among them, the embodiments of the present application can use the PMP algorithm to calculate the SOC value that meets certain optimal conditions under the corresponding driving conditions, and use the SOC value that meets certain optimal conditions as the training samples for training the pre-constructed CNN-LSTM global SOC planning model.

[0168] Step S703: Determine the SOC values that meet certain optimal conditions corresponding to different driving conditions.

[0169] Step S704: Train the pre-constructed CNN-LSTM global SOC planning model.

[0170] Among them, the embodiments of the present application can use the SOC values that meet certain optimal conditions corresponding to different driving conditions to train the pre-constructed CNN-LSTM global SOC planning model, and then obtain the trained CNN-LSTM global SOC planning model.

[0171] Online application stage: Step S705: Obtain actual operating condition data.

[0172] Among them, in the embodiments of the present application, the actual operating condition data of the vehicle under different operating conditions can be collected according to the API map information.

[0173] Step S706: Segment to obtain the actual section operating condition data corresponding to different road sections.

[0174] Step S707: Use the trained CNN-LSTM global SOC planning model to obtain the terminal SOC value.

[0175] Among them, in the embodiments of the present application, the congestion situation , average speed , section length , the starting SOC value at the beginning of the section , total remaining driving mileage and the starting SOC value of the operating condition can be extracted as the input of the trained CNN-LSTM global SOC planning model, and then the starting SOC value of the next section can be obtained by using the trained CNN-LSTM global SOC planning. .

[0176] Step S708: Control the power output according to the SOC value to achieve the energy management of the vehicle.

[0177] Among them, in the embodiments of the present application, the power output can be controlled according to the starting or terminal SOC value to achieve the energy management of the vehicle.

[0178] According to the vehicle energy management method proposed in the embodiments of the present application, based on the obtained actual operating condition data, the road during the vehicle driving process can be segmented, and then the actual section operating condition data corresponding to different road sections can be obtained, and the trained CNN-LSTM global SOC planning model is used to output the terminal SOC value of the vehicle on the corresponding road section. Then, the power output of the vehicle is controlled according to the starting SOC value or the terminal SOC value to achieve the energy management of the vehicle. Through segmented prediction, error accumulation is suppressed, the SOC prediction error is reduced, the SOC prediction value is dynamically adjusted according to the actually received actual operating condition data in real time, the prediction delay is reduced, the real-time requirement is met, and the driving comfort is improved according to different operating conditions, the battery life is extended, and the safety is enhanced. Thus, the problems in the related art, such as being unable to adapt to complex traffic environments and operating conditions, having low intelligence, low energy utilization rate, and being unable to effectively improve fuel economy, are solved.

[0179] Secondly, a vehicle energy management device proposed according to an embodiment of the present application is described with reference to the accompanying drawings.

[0180] Figure 8 Schematic block diagram of an energy management device for a vehicle provided according to another embodiment of the present application.

[0181] As Figure 8 shown, the energy management device 80 of the vehicle is applied to the online application stage of the model. Among them, the device 80 includes: a second acquisition module 801, a second segmentation module 802, a second generation module 803, and a control module 804.

[0182] Among them, the second acquisition module 801 is used to acquire the actual operating condition data of the vehicle under different operating conditions.

[0183] The second segmentation module 802 is used to segment the road during the driving process of the vehicle based on the actual operating condition data, so as to obtain the actual section operating condition data corresponding to different road sections.

[0184] The second generation module 803 is used to input the starting SOC value corresponding to the actual section operating condition data into the trained CNN-LSTM global SOC planning model to obtain the ending SOC value of the vehicle on the corresponding road section, where the trained CNN-LSTM global SOC planning model is trained by the starting SOC value and the actual section operating condition data.

[0185] The control module 804 is used to control the power output of the vehicle according to the starting SOC value or the ending SOC value, so as to realize the energy management of the vehicle.

[0186] It should be noted that the foregoing explanation of the embodiment of the energy management method for the vehicle also applies to the energy management device of the vehicle in this embodiment, and will not be elaborated here.

[0187] The energy management device of the vehicle proposed according to the embodiment of the present application can segment the road during the driving process of the vehicle based on the acquired actual operating condition data, and then obtain the actual section operating condition data corresponding to different road sections, and use it to output the ending SOC value of the vehicle on the corresponding road section by the trained CNN-LSTM global SOC planning model. Furthermore, the power output of the vehicle is controlled according to the starting SOC value or the ending SOC value to realize the energy management of the vehicle. Through segmented prediction, error accumulation is suppressed, the SOC prediction error is reduced, the SOC prediction value is dynamically adjusted according to the actual operating condition data received in real time, the prediction delay is reduced, the real-time requirement is met, and the energy management is adapted to different operating conditions, improving driving comfort, extending battery life, and enhancing safety. Thus, the problems in the related art that it is impossible to adapt to complex traffic environments and operating conditions, the degree of intelligence is not high, the energy utilization rate is low, and the fuel economy cannot be effectively improved are solved.

[0188] Figure 9Schematic structural diagram of a vehicle provided according to an embodiment of the present application. The vehicle may include: A memory 901, a processor 902, and a computer program stored on the memory 901 and executable on the processor 902.

[0189] When the processor 902 executes the program, it implements the energy management method of the vehicle provided in the above embodiment.

[0190] Furthermore, the vehicle further includes: A communication interface 903 for communication between the memory 901 and the processor 902.

[0191] The memory 901 is used to store a computer program executable on the processor 902.

[0192] The memory 901 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory.

[0193] If the memory 901, the processor 902, and the communication interface 903 are implemented independently, the communication interface 903, the memory 901, and the processor 902 may be interconnected through a bus and communicate with each other. The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 9 only a thick line is shown in the figure, but it does not mean that there is only one bus or one type of bus.

[0194] Optionally, in a specific implementation, if the memory 901, the processor 902, and the communication interface 903 are integrated on a chip, the memory 901, the processor 902, and the communication interface 903 may communicate with each other through an internal interface.

[0195] The processor 902 may be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0196] An embodiment of the present application further provides a computer-readable storage medium, on which a computer program is stored. When the program is executed by a processor, the energy management method of the vehicle as described above is implemented.

[0197] An embodiment of the present application further provides a computer program product, including a computer program. When the program is executed, the energy management method of the vehicle as described above is implemented.

[0198] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in any one or N embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0199] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise specifically defined.

[0200] Any process or method description shown in the flowchart or described in other ways herein can be understood as representing a module, segment, or part of code including one or N executable instructions for implementing a customized logic function or process. The scope of the preferred embodiments of the present application includes additional implementations, where the functions may be executed in a substantially simultaneous manner or in a reverse order according to the functions involved, rather than in the order shown or discussed, which should be understood by those skilled in the art to which the embodiments of the present application belong.

[0201] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or in conjunction with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection part (electronic device) having one or N wirings, a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically by optically scanning the paper or other media, followed by editing, interpretation, or other suitable processing as necessary, and then stored in a computer memory.

[0202] It should be understood that various parts of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, the N steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented in hardware, as in another embodiment, it can be implemented by a combination of any one or more of the following techniques well known in the art: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), and the like.

[0203] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the methods of the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0204] In addition, each functional unit in various embodiments of the present application may be integrated into a processing module, may exist separately physically for each unit, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0205] The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disc, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A vehicle energy management method, characterized in that: Applied to the offline model training stage, wherein the method comprises the following steps: Obtaining corresponding operating condition data of at least one vehicle under different operating conditions; Based on the working condition data, the road on which the vehicle is traveling is segmented to obtain the road section working condition data corresponding to different road sections; Based on the road section operating condition data, calculating a state of charge (SOC) value of the at least one vehicle that satisfies a preset optimal condition under a corresponding operating condition; The SOC value that meets the preset optimal conditions and the operating condition data are used to train a pre-constructed convolutional long short-term memory network CNN-LSTM global SOC planning model to obtain a trained CNN-LSTM global SOC planning model, and the SOC value of the vehicle is predicted based on the trained CNN-LSTM global SOC planning model to achieve energy management of the vehicle.

2. The method according to claim 1, characterized in that The step of segmenting the road on which the vehicle is traveling based on the operating condition data includes: Determine at least one of an average vehicle speed, a road section length, and a road section congestion level during the driving process of the vehicle based on the operating condition data; Based on the average vehicle speed and the length of the road section, determining whether the rate of change of the vehicle speed at different positions is greater than a preset value, and / or, based on the congestion level of the road section and the length of the road section, determining whether the congestion level of the road section meets a preset condition; If the vehicle speed change rate is greater than the preset value, the corresponding road is segmented according to the average vehicle speed and the road section length, and / or, if the road section congestion level meets the preset condition, the corresponding road is segmented according to the road section congestion level and the road section length; If the vehicle speed change rate is less than or equal to the preset value, the corresponding road is not segmented, and / or if the congestion level of the road section does not meet the preset condition, the corresponding road is not segmented.

3. The method according to claim 1, characterized in that The obtaining of operating condition data corresponding to at least one vehicle under different operating conditions includes: Acquiring initial driving data of the at least one vehicle under corresponding working conditions; Acquiring initial road data of the at least one vehicle under corresponding working conditions; Processing the initial driving data and the initial road data respectively to obtain corresponding driving data and road data; The operating condition data is obtained based on the driving data and the road data.

4. The method according to claim 1, characterized in that Before using the SOC value that meets the preset optimal condition and the operating condition data to train the pre-built CNN-LSTM global SOC planning model, the method further includes: Constructing an input feature matrix of a CNN-LSTM global SOC planning model using the SOC value that meets the preset optimal condition and the operating condition data; Based on the input feature matrix, determining the CNN feature extraction layer of the CNN-LSTM global SOC planning model; Using the CNN feature extraction layer to obtain an output one-dimensional vector corresponding to the input feature matrix; Based on the output one-dimensional vector, determining the LSTM network of the CNN-LSTM global SOC planning model; The CNN-LSTM global SOC planning model is constructed using the CNN feature extraction layer and the LSTM network.

5. The method according to claim 4, characterized in that The step of determining the LSTM network of the CNN-LSTM global SOC planning model based on the output one-dimensional vector includes: Based on the output one-dimensional vector, determining an input gate of the LSTM network; Obtain the cell state and hidden state of the LSTM network; Determine a forget gate of the LSTM network based on the input gate and the hidden state; Determine an output gate of the LSTM network based on the input gate, the cell state, the hidden state and the forget gate; The LSTM network is obtained by using the input gate, the cell state, the hidden state, the forget gate and the output gate.

6. The method according to claim 1, characterized in that The method of training a pre-built CNN-LSTM global SOC planning model using the SOC value that meets the preset optimal condition and the operating condition data to obtain a trained CNN-LSTM global SOC planning model includes: Obtain the first-order moment estimation exponential decay rate, the second-order moment estimation exponential decay rate and the learning rate of the adaptive moment estimation optimizer Adam optimizer corresponding to the training of the pre-built CNN-LSTM global SOC planning model; Based on the first-order moment estimated exponential decay rate, the second-order moment estimated exponential decay rate and the learning rate, the pre-constructed CNN-LSTM global SOC planning model is trained until the pre-constructed CNN-LSTM global SOC planning model meets the preset training conditions to obtain the trained CNN-LSTM global SOC planning model.

7. The method according to claim 1, characterized in that The predicting the SOC value of the vehicle based on the trained CNN-LSTM global SOC planning model to achieve energy management of the vehicle includes: Based on the SOC value, determining the energy consumption per unit distance of the vehicle on the corresponding road section; Using the road section operating condition data, the SOC value and the energy consumption per unit distance, construct an energy management database corresponding to the vehicle; The energy management database is used to generate an energy management strategy for the vehicle to implement energy management of the vehicle.

8. A vehicle energy management method, characterized in that: The vehicle energy management method according to any one of claims 1 to 7 is applied to the model online application stage, wherein the method comprises the following steps: Obtain the actual working condition data of the vehicle under different working conditions; Based on the actual working condition data, the road on which the vehicle is traveling is segmented to obtain actual road section working condition data corresponding to different road sections; Inputting the starting SOC value corresponding to the actual road section operating condition data into the trained CNN-LSTM global SOC planning model to obtain the ending SOC value of the vehicle on the corresponding road section, wherein the trained CNN-LSTM global SOC planning model is trained by the starting SOC value and the actual road section operating condition data; The power output of the vehicle is controlled according to the starting SOC value or the ending SOC value to achieve energy management of the vehicle.

9. A vehicle, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the energy management method for a vehicle as described in any one of claims 1 to 7 or the energy management method for a vehicle as described in claim 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the energy management method for a vehicle as described in any one of claims 1 to 7 or the energy management method for a vehicle as described in claim 8.

Images