Extended-range vehicle energy management method, device, electronic device and storage medium

Abstract

The present application provides an energy management method, device, electronic device and storage medium for an extended-range vehicle. The method includes: obtaining the historical speed information of the vehicle, processing the historical speed information using a pre-built speed prediction model to obtain the future predicted speed; calculating the vehicle power demand, and planning the change trajectory of the battery state of charge according to the vehicle power demand and the expected driving distance to obtain a reference trajectory of the battery state of charge; according to the future predicted speed and the reference trajectory of the battery state of charge, converting the preset multi-objective optimization control objective function into a nonlinear programming problem, and solving the nonlinear programming problem using a genetic algorithm that integrates an elite strategy to obtain an optimal energy management strategy, and using the optimal energy management strategy to perform energy management on the extended-range vehicle. The present application improves the accuracy of speed prediction, improves the efficiency of energy management of extended-range vehicles, and reduces the complexity and cost of the solution.

Description

Technical Field

[0001] The present application relates to the field of new energy vehicle technology, and in particular to an energy management method, device, electronic device and storage medium for an extended-range vehicle. Background Art

[0002] With the advancement of technology, extended-range vehicles are increasingly used in modern transportation systems. Such vehicles can operate in multiple operating modes, and the energy usage and demand in each mode are different, which increases the complexity of energy management. In order to optimize energy consumption and meet driving needs, an efficient energy management strategy is needed to allocate the energy of the vehicle.

[0003] With the help of intelligent transportation systems, vehicles can receive global positioning information, but existing energy management systems have limitations in practical applications. Most existing methods rely on long-term SOC planning and short-term state prediction. These systems often require global driving information to support energy management, but it is very difficult to obtain accurate global driving information under changing real-time driving conditions.

[0004] In addition, existing speed prediction methods are mainly based on networks such as LSTM or Bi-LSTM, which can only process historical speed data and fail to fully utilize the deep topological relationship between speed data. Although graph convolutional networks (GCNs) have been proven to effectively process graph structured data and mine deep relationships between nodes, how to effectively integrate GCNs into speed prediction and energy management strategies to achieve optimization of battery life and fuel economy remains a technical challenge that needs to be solved.

[0005] At present, the global optimization method for battery life and fuel economy is feasible in theory, but it is not easy to implement in practice. Therefore, it is urgent to develop a new energy management method. Summary of the invention

[0006] In view of this, the embodiments of the present application provide an extended-range vehicle energy management method, device, electronic device and storage medium to solve the problems of low accuracy of vehicle speed prediction, reduced efficiency of extended-range vehicle energy management, complex solutions and high costs in the prior art.

[0007] In a first aspect of an embodiment of the present application, a method for energy management of an extended-range vehicle is provided, comprising: obtaining historical speed information of the vehicle, processing the historical speed information using a pre-built speed prediction model to obtain a future predicted speed, wherein the speed prediction model processes spatial feature interactions through a graph convolutional network, and processes time series predictions through a long short-term memory network; calculating the vehicle power demand, and planning the change trajectory of the battery state of charge according to the vehicle power demand and the expected driving distance to obtain a battery state of charge reference trajectory; according to the future predicted speed and the battery state of charge reference trajectory, converting a preset multi-objective optimization control objective function into a nonlinear programming problem, and solving the nonlinear programming problem using a genetic algorithm that integrates an elite strategy to obtain an optimal energy management strategy, and using the optimal energy management strategy to perform energy management on the extended-range vehicle.

[0008] According to a second aspect of an embodiment of the present application, there is provided an energy management device for an extended-range vehicle, comprising: a processing module configured to obtain historical speed information of the vehicle, and process the historical speed information using a pre-built speed prediction model to obtain a future predicted speed, wherein the speed prediction model processes spatial feature interactions through a graph convolutional network, and processes time series predictions through a long short-term memory network; a planning module configured to calculate the vehicle power demand, and plan the change trajectory of the battery state of charge according to the vehicle power demand and the expected driving distance, and obtain a battery state of charge reference trajectory; a solving module configured to convert a preset multi-objective optimization control objective function into a nonlinear planning problem according to the future predicted speed and the battery state of charge reference trajectory, and solve the nonlinear planning problem using a genetic algorithm that integrates an elite strategy to obtain an optimal energy management strategy, and use the optimal energy management strategy to perform energy management on the extended-range vehicle.

[0009] According to a third aspect of an embodiment of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the above method are implemented when the processor executes the computer program.

[0010] According to a fourth aspect of an embodiment of the present application, a computer-readable storage medium is provided, which stores a computer program, and when the computer program is executed by a processor, the steps of the above method are implemented.

[0011] At least one of the above technical solutions adopted in the embodiments of the present application can achieve the following beneficial effects:

[0012] By obtaining the historical speed information of the vehicle, the historical speed information is processed using a pre-built speed prediction model to obtain the future predicted speed, wherein the speed prediction model processes spatial feature interactions through a graph convolutional network and processes time series predictions through a long short-term memory network; the vehicle power demand is calculated, and the battery state of charge change trajectory is planned based on the vehicle power demand and the expected driving distance to obtain a battery state of charge reference trajectory; based on the future predicted speed and the battery state of charge reference trajectory, the preset multi-objective optimization control objective function is converted into a nonlinear programming problem, and the nonlinear programming problem is solved using a genetic algorithm that integrates an elite strategy to obtain the optimal energy management strategy, and the optimal energy management strategy is used to manage the energy of the extended-range vehicle. This application improves the accuracy of speed prediction, improves the efficiency of energy management of extended-range vehicles, and reduces the complexity and cost of the solution. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0014] Figure 1 It is a schematic diagram of the overall implementation process of the multi-objective optimization extended-range vehicle energy management method based on double-layer prediction provided in an embodiment of the present application;

[0015] Figure 2 It is a flow chart of the energy management method of the extended-range vehicle provided in the embodiment of the present application;

[0016] Figure 3 It is a schematic diagram of the construction process of the graph structure provided in the embodiment of the present application;

[0017] Figure 4 It is a structural diagram of a multi-feature interaction mechanism provided in an embodiment of the present application;

[0018] Figure 5 It is a schematic diagram of the LSTM vehicle speed prediction process with multiple information inputs provided in an embodiment of the present application;

[0019] Figure 6 It is a flow chart of a genetic algorithm integrating an elite strategy provided in an embodiment of the present application;

[0020] Figure 7 It is a structural schematic diagram of an energy management device for an extended-range vehicle provided in an embodiment of the present application;

[0021] Figure 8 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0022] In the following description, specific details such as specific system structures, technologies, etc. are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present application. However, it should be clear to those skilled in the art that the present application may also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present application.

[0023] In recent years, with the development of extended-range vehicles, the nature of their operation in different operating modes has led to the complexity of selecting the operating mode and controlling energy distribution. In order to achieve different optimization goals (such as reducing energy consumption) while meeting driving needs, the vehicle controller needs to adopt an efficient energy management strategy to reasonably distribute energy. The widespread application of technologies such as global positioning system (GPS), intelligent transportation system (ITS), geographic information system (GIS), and Internet of Vehicles (IOV) enables vehicles to obtain global driving information during operation. Therefore, more and more energy management studies consider taking driving information into consideration.

[0024] Most existing methods achieve energy management by combining long-term state of charge (SOC) planning and short-term state prediction using a hierarchical predictive EMS with traffic awareness. However, this hierarchical energy management relies heavily on the designed global SOC reference trajectory, and it is quite difficult to accurately obtain global driving information under the complex changes of real-time driving conditions. Therefore, in further research, it is necessary to seriously consider how to obtain the SOC reference trajectory without relying on global driving information. With the development of model predictive control theory, energy management research based on model predictive control has received widespread attention.

[0025] Vehicle speed prediction is one of the keys to model predictive control. However, existing vehicle speed prediction methods only use networks such as LSTM and Bi-LSTM to recursively mine historical vehicle speed information, and cannot mine the hidden topological relationship between multiple features of vehicle speed. To improve this, GCN is a neural network model that can process graph-structured data, which can effectively learn and mine the relationship between nodes in the graph. The vehicle speed prediction model based on GCN can predict future power demand and solve short-term energy management problems through optimization algorithms to find a near-optimal solution. In addition, for extended-range vehicles, since batteries are expensive and have a relatively short service life, it is equally important to extend the operating durability of batteries by rationally designing EMS. However, the current global optimization method that considers battery life economy optimization is only feasible in theory. Therefore, it is necessary to extensively study how to design a multi-objective environmental management system that considers battery life and fuel economy without considering global driving information.

[0026] According to the above existing technical solutions, the existing technology has at least the following problems:

[0027] 1) SOC global trajectory planning problem: SOC (battery state of charge) refers to the percentage of remaining energy in the battery. Traditionally, effective energy management requires pre-planning of the SOC trajectory for the entire trip, which is difficult under constantly changing real-time conditions.

[0028] 2) Vehicle speed multi-feature relationship mining problem: Vehicle speed has a complex relationship with many other factors such as acceleration, road conditions, traffic conditions, etc. If this information can be fully utilized, it will help to better predict vehicle speed and thus conduct more effective energy management.

[0029] 3) Multi-objective optimization problem: Battery life and fuel economy are usually conflicting objectives, and optimizing these two objectives requires complex decision-making strategies.

[0030] In view of the problems existing in the prior art, the present application aims to provide an energy management method for extended-range vehicles based on double-layer prediction and multi-objective optimization. This application proposes an energy management method for extended-range vehicles based on double-layer prediction and multi-objective optimization to address the problems in existing methods that the global trajectory of SOC is difficult to plan, the hidden topological relationship between multiple features of vehicle speed cannot be mined, and the multi-objective optimization of battery life and fuel economy is difficult to optimize. First, a multi-feature interactive vehicle speed prediction module is constructed, and the vehicle speed information in the prediction layer is estimated by combining GCN and LSTM; then a global planning method of the state of charge (SOC) based on the driving distance is used. According to the predicted vehicle speed and SOC, a multi-objective optimization model is input, and the direct point allocation method is combined with a genetic algorithm that integrates the elite strategy to optimize the battery life economy and fuel economy within the prediction range.

[0031] Before describing the embodiments of the present application in detail, the overall implementation process of the technical solution of the present application is first summarized in combination with the accompanying drawings and specific embodiments. Figure 1 This is a schematic diagram of the overall implementation process of the multi-objective optimization extended-range vehicle energy management method based on double-layer prediction provided in the embodiment of the present application. Figure 1 As shown, the multi-objective optimization extended-range vehicle energy management method based on double-layer prediction may specifically include the following contents:

[0032] The overall implementation process of the technical solution of this application includes: in the first prediction layer, a multi-feature interactive speed prediction module is constructed, and the multi-features of historical speed are interactively extracted based on GCN, and then the future speed information is accurately predicted through LSTM rolling prediction; then, a SOC planning method based on driving distance without considering global driving information is proposed. Since this method does not rely on global driving information, it can perform SOC planning in a real-time environment and quickly calculate the appropriate charging or discharging strategy to meet driving needs. In the second multi-objective optimization layer, according to the SOC reference trajectory and speed information of the prediction layer, the direct point matching method is combined with the genetic algorithm fused with the elite strategy, and the optimization control problem is converted into a nonlinear planning problem and then solved by the genetic algorithm fused with the elite strategy to achieve optimal control within the prediction range, thereby improving fuel economy and battery life.

[0033] It can be seen that the present application proposes an energy management method for extended-range vehicles based on double-layer prediction and multi-objective optimization. In the first prediction layer, a multi-feature interactive vehicle speed prediction module is constructed. The corresponding relationship between the historical vehicle speed and acceleration is established through the multi-feature interaction mechanism. The features are interactively mined through information transmission and information selection. Then, the future vehicle speed is accurately predicted through the LSTM with multi-information input. Then, a global planning method is used for the state of charge (SOC) based on the driving distance, and SOC planning is performed in a real-time environment. In the second multi-objective optimization layer, according to the SOC reference trajectory and vehicle speed information of the prediction layer, the direct point matching method is combined with the genetic algorithm fused with the elite strategy. The optimization control problem is converted into a nonlinear planning problem and then solved by the genetic algorithm fused with the elite strategy to achieve optimal control within the prediction range, thereby improving fuel economy and battery life.

[0034] The contents of the technical solution of the present application are described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0035] Figure 2 It is a flow chart of the extended-range vehicle energy management method provided in an embodiment of the present application. Figure 2 The energy management method of the extended-range vehicle can be executed by the vehicle energy management system. Figure 2 As shown, the extended-range vehicle energy management method may specifically include:

[0036] S201, obtaining historical speed information of the vehicle, and processing the historical speed information using a pre-built speed prediction model to obtain a future predicted speed, wherein the speed prediction model processes spatial feature interactions through a graph convolutional network and processes time series prediction through a long short-term memory network;

[0037] S202, calculating the power demand of the vehicle, and planning the change trajectory of the battery state of charge according to the power demand of the vehicle and the expected driving distance, to obtain a reference trajectory of the battery state of charge;

[0038] S203, based on the future predicted vehicle speed and the reference trajectory of the battery state of charge, the preset multi-objective optimization control objective function is converted into a nonlinear programming problem, and the nonlinear programming problem is solved using a genetic algorithm that integrates an elite strategy to obtain an optimal energy management strategy, which is used to manage the energy of the extended-range vehicle.

[0039] In some embodiments, before processing the historical vehicle speed information using the pre-built vehicle speed prediction model, the method further includes:

[0040] Obtain historical vehicle speed vectors and historical acceleration vectors, and divide the time window into multiple time segments. Use the historical vehicle speed vector or historical acceleration vector corresponding to each time segment as a node, and the relationship between the nodes as an edge. Use the nodes and the edges between the nodes to construct a graph structure, where the graph structure includes a vehicle speed graph structure and an acceleration graph structure.

[0041] Specifically, in the first prediction layer, the embodiment of the present application constructs a multi-feature interactive vehicle speed prediction model, the main purpose of which is to predict the vehicle dynamics in the future based on historical vehicle speed information and current vehicle status. The vehicle speed prediction model uses historical vehicle speed (m / s), acceleration (m / s 2 ) information to improve the accuracy of future vehicle speed prediction. First, select the historical vehicle speed vector V1 = [V k-h ,V k-h+1 ,…,V k ], and then obtain the corresponding acceleration vector V2 = [V k2-h ,V k2-h+1 ,…,V k2 ], then segment V1 and V2, with T = 2s as a segment, and use the graph convolutional network to interactively extract each segment. Time segment segmentation means further subdividing the time window into smaller time segments, and the length of each time segment is T = 2 seconds.

[0042] Furthermore, in GCN (graph convolutional network), two graph structures must first be constructed. Figure 3 is a schematic diagram of the construction process of the graph structure provided in the embodiment of the present application, such as Figure 3 As shown, the process of constructing the vehicle speed graph structure and the acceleration graph structure in the embodiment of the present application may include the following contents:

[0043] For each time segment, the data points of vehicle speed and acceleration are regarded as nodes in the graph. The characteristics of the node are determined by the vehicle speed or acceleration information in the corresponding time segment. The graph structures G1 and G2 correspond to vehicle speed and acceleration respectively, so this embodiment constructs two graphs with the same structure respectively. Each graph consists of nodes (i.e., time segments) and edges (relationships between nodes).

[0044] In practical applications, cosine similarity is used to measure the similarity between time segments (nodes). This measure reflects the similarity of the data change trends in two time segments. If the similarity is high, an edge will be established between the two time segments in the graph. The adjacency matrix A is a matrix that represents the relationship between nodes. If there is a connection between two nodes (that is, the speed or acceleration time segments between them are similar), the corresponding position in the adjacency matrix will have a non-zero value.

[0045] In a specific example, if Figure 3 As shown in the figure, for an undirected graph G = (V, E, A), it is composed of a set of non-empty vertices and the edges between them. Where V represents the set of nodes in the graph, E is the set of edges, and A represents the adjacency matrix of the graph G. First, V1 is constructed as a graph, and each segment is regarded as a node. The node feature is the node feature. Then, the node connection is constructed according to the cosine similarity, and the graph structure of V1 is obtained as G1 = (V 1T , E1, A1), the graph structure of V2 can be the same as that of V1, G2 = (V 1T , E1, A1), the acceleration of the node feature in each segment, by maintaining the same graph structure, a corresponding relationship can be established between V1 and V2, so as to facilitate feature interaction and information transfer in the graph convolutional network.

[0046] In some embodiments, the historical vehicle speed information is processed using a pre-built vehicle speed prediction model to obtain a future predicted vehicle speed, including:

[0047] Run a graph convolutional network on the graph structure, use the edges in the graph structure to transfer information, and use the graph convolutional network to interactively learn the features between different nodes through the connection relationship between nodes, so as to update the features of each node, where the updated node incorporates the information of adjacent nodes;

[0048] The updated nodes are used as inputs of the long short-term memory network, so that the long short-term memory network predicts the vehicle speed within a preset time period in the future using a rolling window method based on the information contained in the updated nodes.

[0049] Specifically, in this embodiment, in order to improve the accuracy and efficiency of extended-range vehicle energy management, a vehicle speed prediction model that comprehensively utilizes graph convolutional networks (GCN) and long short-term memory networks (LSTM) is provided. Among them, GCN is used to capture the spatial relationship between vehicle speed data, such as how historical vehicle speed data influence each other; LSTM is used to process time series data to predict future vehicle speeds. This model aims to fully capture the complex relationship between historical vehicle speed and acceleration data, and accurately predict vehicle speeds in the future, thereby guiding the energy management system to effectively allocate energy.

[0050] First, the historical data of vehicle speed and acceleration over a period of time are collected and divided into multiple time segments according to the time window. The speed and acceleration data points in each time segment are regarded as a node in the graph structure, and the connection relationship (edge) between nodes is established based on the similarity or physical proximity of these data points. The graph structure aims to reveal the intrinsic correlation between the changes in vehicle speed and acceleration between different time segments.

[0051] Then, GCN is deployed to process graph structure data. The core of GCN is the ability to transmit information through the edges between nodes, allowing the features of each node to interact and update. This processing makes the features of each node not only include the speed and acceleration information within the time segment, but also integrates the relevant information of the surrounding time segments, reflecting the complex dynamic changes caused by factors such as road conditions and traffic flow.

[0052] Finally, the updated graph structure node features are then used as input to the LSTM network. The LSTM network is particularly suitable for processing time series data and can learn and remember long-term dependencies in time series. Through a rolling window, the LSTM network combines the spatial and temporal features from the GCN to predict the future vehicle speed. The speed prediction result is used to guide the energy management system to plan the SOC, ensuring that energy efficiency is maximized and battery life is extended while providing sufficient power.

[0053] The vehicle speed prediction model provided in this embodiment combines the capabilities of GCN in spatial relationship learning and the advantages of LSTM in time series prediction, thereby providing a more accurate and reliable energy management method for extended-range vehicles. Through this model, the energy management system can more flexibly adapt to the ever-changing driving environment and achieve multi-objective optimization, such as reducing energy consumption and improving battery life.

[0054] In the above process, the combination of GCN and LSTM enables the vehicle speed prediction model to take into account both the temporal dependency and spatial topological dependency of time series data. In this way, the vehicle speed prediction model can not only learn the changes in vehicle speed over time, but also understand the complex dependencies between different time segments, thereby improving the prediction accuracy. This is crucial for energy management systems because it can adjust the battery charging and discharging strategy based on more accurate vehicle speed predictions to optimize energy use, extend battery life, and improve fuel efficiency.

[0055] In some embodiments, the graph convolutional network is used to interactively learn the features between different nodes through the connection relationship between the nodes so as to update the features of each node, including:

[0056] Based on the information propagation mechanism in the graph convolutional network, the features of each node are updated according to the edges defined in the graph structure;

[0057] During the information propagation process, the interaction ratio between the vehicle speed feature and the acceleration feature is adjusted using the preset hyperparameters to balance the contribution of feature updates;

[0058] Perform information selection operations by weighted accumulation of node features updated through different information propagation steps, and use the parameter matrix to filter the information propagated in each layer, and use the final filtered features as the input of the long short-term memory network.

[0059] Specifically, GCN is run on the constructed graph structure to learn the characteristics of each node (time segment), that is, the relationship between vehicle speed and acceleration, through the connection relationship between nodes. The results of GCN processing are input into the LSTM network, which uses this information to predict future vehicle speed because it can understand the pattern of time series data.

[0060] The multi-feature interaction mechanism proposed in this embodiment and the LSTM vehicle speed prediction process based on multi-information input will be described in detail below with reference to the accompanying drawings and specific embodiments. Figure 4 is a schematic diagram of the structure of the multi-feature interaction mechanism provided in the embodiment of the present application, Figure 5 Schematic diagram of the LSTM vehicle speed prediction process with multiple information inputs provided in the embodiment of the present application. Figures 4 to 5 As shown, the multi-feature interaction mechanism of the embodiment of the present application and the process of predicting the future vehicle speed using the LSTM network may specifically include the following contents:

[0061] The embodiment of the present application proposes a multi-feature interaction mechanism to process the information flow on the spatially dependent nodes. The multi-feature interaction mechanism includes two steps: interactive information propagation and information selection. The calculation process is as follows:

[0062]

[0063] Among them, β is a hyperparameter that controls the interaction ratio that preserves the acceleration characteristics; K represents the propagation depth; I is the degree matrix; H2 (k-1) Represents the acceleration characteristics of the previous (k-1) layer input. The information selection step is defined as follows:

[0064]

[0065] Among them, H out Represents the output hidden state of the current layer; W (k) Represents the corresponding weight matrix.

[0066] exist Figure 4 In this embodiment, the information propagation step and the information selection step in the proposed multi-feature interaction mechanism are demonstrated. It first propagates the interaction information horizontally and then selects the information vertically. The information propagation step recursively propagates the node information along the given graph structure. In the extreme case where there is no spatial dependency, aggregating neighborhood information will add useless noise to each node. Therefore, the information selection step is introduced to filter out the unimportant information generated by each hop, and the parameter matrix W in the above formula (2) is (k) as a feature selector.

[0067] Furthermore, the data Hout processed by the multi-feature interaction mechanism is used as the input of the LSTM network, and the LSTM network is used to predict the future vehicle speed. LSTM uses its ability to process time series data, combined with historical vehicle speed and acceleration information, to calculate the change in vehicle speed over a period of time in the future. The LSTM network usually makes predictions through a rolling window method, so it gradually moves the time window and continuously updates its prediction of future vehicle speed. In practical applications, the time step of historical data can be selected as 2 seconds, which means that the system will analyze the data every 2 seconds and make a prediction of the vehicle speed.

[0068] For example, in one example, the LSTM with multiple information inputs can be used to accurately predict future vehicle speed information. First, the fusion information H of the historical vehicle speed and acceleration information obtained by the multi-feature fusion mechanism is used. out , historical speed, and historical acceleration are input into the historical time domain, and the speed change in the next p seconds is calculated. Then, the rolling window is used for future speed prediction, so the process of LSTM predicting vehicle speed is achieved through rolling prediction.

[0069] According to the technical solution provided by the above embodiment, the purpose of the entire multi-feature interaction mechanism is to extract and combine relevant information in the historical vehicle speed and acceleration data to better understand the relationship between these features and predict future vehicle speeds. Through the information propagation and selection steps, the model can focus more on those features that have the greatest impact on the prediction results while removing unimportant noise. This mechanism is of great significance in the fields of intelligent transportation and autonomous driving, and it can help improve the speed and accuracy of the vehicle control system's response to future situations.

[0070] In some embodiments, the vehicle power demand is calculated, and the battery state of charge change trajectory is planned according to the vehicle power demand and the expected driving distance, including:

[0071] The vehicle power demand is calculated using a predetermined simplified vehicle power demand model to obtain the total power required by the vehicle, and the total power required by the vehicle is associated with the driving distance so that the total power required by the vehicle is linearly correlated with the driving distance. Based on the linear relationship between the total power required by the vehicle and the driving distance, a change trajectory of the battery state of charge is established; wherein the simplified vehicle power demand model calculates the total power required by the vehicle based on the slope resistance, rolling resistance, air resistance and inertial resistance of the vehicle.

[0072] Specifically, the embodiment of the present application also describes in detail how to plan the change trajectory of the battery state of charge (SOC) based on the driving distance. In practical applications, the planning method of the battery state of charge change trajectory is as follows:

[0073] The SOC planning method based on driving distance does not rely on global vehicle speed information, but plans the SOC reference trajectory according to the expected driving distance. Compared with the traditional method, this method is fast and practical because the driving distance is relatively stable and can be obtained through GPS and other methods, and is not easily affected by external factors.

[0074] First, the power of the vehicle can be expressed as the sum of slope resistance, rolling resistance, air resistance, and inertial resistance. The power required by the Pwheel wheel can be calculated as:

[0075]

[0076] Among them, C D represents the windward drag coefficient, A represents the windward area, v(t) represents the vehicle speed, and f r represents the rolling resistance coefficient, m represents the vehicle mass, g represents the acceleration due to gravity, δ represents the rotational mass coefficient, a(t) represents the vehicle acceleration, and ρ represents the air density.

[0077] Usually, it is assumed that the end SOC is its lower limit. The task of the energy management strategy is to find a trajectory of SOC decline, starting from the initial value and ending at the final value. During driving, it is simply assumed that when the vehicle speed is low, the air resistance is small and can be ignored. At the same time, the average acceleration is taken to simplify the power calculation. The power required by the vehicle during driving can be expressed according to formula (3) as:

[0078] W=[(mgf r cosθ+mg sinθ)+δma(t)]×S (4)

[0079] Under this assumption, the vehicle's required power increases linearly with the driving distance; therefore, the change in state of charge is approximately a linear function of the driving distance, which can be expressed as:

[0080]

[0081] Where V(k) represents the speed at time k; S(k) represents the distance traveled at time k; S(k+p) represents the distance traveled at time k+p; SOC0 represents the initial remaining battery capacity; S all Indicates the total trip distance obtained at the beginning of each trip; SOC min It represents the minimum acceptable capacity of the battery, that is, the minimum state of charge that the battery needs to maintain; SOC(K+p) represents the SOC end value at time k+p when the prediction horizon ends, so that the SOC reference trajectory can be quickly determined.

[0082] In practical applications, in order to simplify calculations, it can be assumed that air resistance is small at low speeds and can be ignored. In addition, the average value of acceleration can be taken to simplify the power calculation. Based on the above simplifications, the total power required by the vehicle is approximately linearly related to the driving distance, as shown in formula (4). Through the relationship between power demand and driving distance, the change trajectory of SOC can be established. If the total driving distance (S all ) and the current driving distance (S(k)), as well as the initial SOC (SOC0) and the minimum allowed SOC (SOC min ), the SOC value at a certain moment in the future can be estimated, as shown in formula (5). By real-time monitoring and predicting the vehicle speed, the SOC planning can be adjusted in real time according to the actual driving distance and known road conditions (such as slope).

[0083] Through the method of the above embodiment, the system can adjust the battery charging and discharging strategy based on the actual driving distance and predicted future driving conditions instead of relying on the global vehicle speed information. The advantages of this method are fast and practical, and it can adapt to changing driving conditions and update the SOC trajectory in real time to ensure that the battery operates at an ideal state of charge, thereby improving energy efficiency and battery life. This SOC planning method based on driving distance can be executed quickly in actual operation because it does not require complex calculations, but only estimates based on driving distance and basic physical formulas.

[0084] In some embodiments, according to the future predicted vehicle speed and the battery state of charge reference trajectory, the preset multi-objective optimization control objective function is converted into a nonlinear programming problem, including:

[0085] The battery state of charge is taken as the state variable, the range extender output current is taken as the control variable, constraints are set for the state variables and control variables, and the objective function of multi-objective optimization control is constructed based on the state variables, control variables and constraints;

[0086] The objective function of multi-objective optimization control is transformed into a nonlinear programming problem using a direct collocation method, wherein the direct collocation method includes using piecewise polynomial equations to approximate the motion trajectory of state variables so as to discretize the objective function of multi-objective optimization control.

[0087] Specifically, the second layer of the system is the SOC planning and multi-objective optimization control layer based on driving distance. This layer focuses on how to use the vehicle speed information predicted by the first layer and other input information to optimize the battery's SOC trajectory and fuel usage. Among them, the multi-objective optimization adopts the direct point matching method and the genetic algorithm fused with the elite strategy. The direct point matching method decomposes the optimization problem into multiple time points, and optimizes the SOC and other control variables at these time points. The establishment of the objective function and the process of converting the objective function into a nonlinear programming problem are described in detail below in conjunction with specific embodiments, which may specifically include the following contents:

[0088] In the second multi-objective optimization layer, according to the SOC reference trajectory and vehicle speed information in the prediction layer, the optimization control problem is transformed into a nonlinear programming problem through the direct point matching method. The state of charge (SOC) is regarded as the state q(t), and the output current I F is regarded as the control variable u(t). The objective function of the prediction layer is:

[0089]

[0090] Among them, m I (t) represents the output current of the range extender at time t; m life(t) represents the battery life attenuation at time t. In addition, the system control variables and state variables should be subject to the following constraints:

[0091]

[0092] Among them, SOC min =0.3, SOC max =0.9,I Fmin =2A,I Fmax =45A.

[0093] Then, the motion trajectory of the state variables is approximated by piecewise polynomial equations, and the original problem is transformed into a nonlinear programming problem with multiple optimization variables and constraints.

[0094] The time interval is [t0, t q ] needs to be evenly divided into n subintervals in the prediction horizon, each subinterval has n+1 nodes. Therefore, the duration of each segment Δt is equal to (t q -t0) / (n+1). Time t i The state variables and control variables are q i =q(t i ) and u i =u(t i ). The optimization parameters are determined by all state variables q i and the control variable u i Therefore, the decision vector of the nonlinear programming problem can be expressed as:

[0095]

[0096] Then, we find that if we choose q i ,q i-1 、u i and u i-1 If the appropriate value of makes the deviation small enough, the polynomial at the midpoint can be approximated as the system's differential equation of motion. In this way, the optimal control problem can be transformed into a nonlinear sequence optimization problem, which requires the optimization of the state variables and control quantities of n+1 nodes.

[0097] When Δt, equation (7) can be transformed into equation (10) by direct collocation method:

[0098]

[0099] Among them, the parameters that need to be optimized include:

[0100] Similarly, the equality constraints for nonlinear programming problems can be expressed as:

[0101]

[0102] The inequality constraints of the optimization problem are summarized as follows:

[0103]

[0104] The objective function and constraints of multi-objective optimization can be transformed into a nonlinear programming problem, where the objective function is: minJ = O(Z).

[0105] According to the above embodiments, the present application adopts a piecewise polynomial approximation method, using piecewise polynomial equations to approximate the trajectory of the state variables, which allows the optimization problem to be discretized and solved by a genetic algorithm. Discretization and the construction of a decision vector divide the prediction time interval into several sub-intervals, each of which corresponds to the value of a state variable and a control variable, forming a decision vector Z, which contains all the variables that need to be optimized, thereby converting the optimal control problem into a nonlinear sequence optimization problem, in which the state variables and control quantities at multiple moments are optimized.

[0106] In some embodiments, a genetic algorithm integrated with an elite strategy is used to solve a nonlinear programming problem to obtain an optimal energy management strategy, including:

[0107] Initialize the population and generate a series of initial solutions, where each initial solution consists of the control variables and state variables to be optimized;

[0108] Calculate individual fitness values ​​and apply the fast non-dominated sorting algorithm to stratify individuals in the population according to dominance relationships and determine the dominance level and crowding degree of each individual;

[0109] Combining parent and offspring individuals, selecting individuals to form a new population based on dominance rank and crowding;

[0110] New individuals are generated through selection, crossover and mutation operations, and the generation operations of new populations and individuals are repeated until the preset iteration termination conditions are reached. Finally, the individual with the highest fitness is selected as the optimal solution, which contains the optimal energy management strategy under the predetermined constraints.

[0111] Specifically, the genetic algorithm that integrates the elite strategy is a search algorithm that searches for the optimal solution by simulating natural selection and genetic principles. The integration of the elite strategy means that the algorithm not only considers the current optimal solution, but also combines excellent solutions within a certain range, which helps to find a more global optimal solution and avoid falling into a local optimum. The vehicle speed and SOC trajectory obtained by the prediction layer are used as input, and after nonlinear programming and genetic algorithm processing, an optimal energy management strategy is obtained while maintaining a balance between fuel economy and battery life. The implementation process of the genetic algorithm that integrates the elite strategy is described in detail below in conjunction with the accompanying drawings and specific embodiments. Figure 6is a flow chart of a genetic algorithm integrating an elite strategy provided in an embodiment of the present application. Figure 6 As shown, the embodiment of the present application uses a genetic algorithm integrated with an elite strategy to solve a nonlinear programming problem, and the process of obtaining an optimal energy management strategy may include the following:

[0112] This example will use the genetic algorithm (NSGA) method integrated with the elite strategy to solve the nonlinear programming problem. The elite strategy is to obtain the dominance level Z by mixing the parent population with its offspring population and performing a fast non-dominated sort. i , and introduce the crowding comparison operator, and convert the crowding degree d i As a comparison criterion between individuals in a population, by comparing Z i d i Screen out outstanding individuals and maintain population diversity.

[0113] Further, the NSGA model can be simplified as follows: NSGA = {C, J, POP, GEN, M, N, V, A, B, C}; wherein C is an individual encoding method, in which the power thresholds X1, X2, X3, X4, SOC thresholds, etc. of the variables to be optimized are converted into character string genes in a binary manner, and are concatenated into a chromosome, that is, a vector X in the optimization solution space represents a chromosome encoding; J is an individual fitness evaluation function, and this embodiment uses minJ = O (Z) as the objective function; POP is an initial population, and a uniform random method is used to generate the initial population; M is the number of objective functions, and this embodiment sets 1:N as the number of populations; V is a dimension, indicating the number of decision variables, and a value is randomly selected from the maximum and minimum values ​​as each decision variable; A is a selection operator; B is a mutation operator, and a basic bit mutation operator is used; C is an operation stop condition algebra, and the operation is set to stop when the maximum number of iterations is reached.

[0114] Through the method of the above embodiment, the process will iteratively search for the optimal solution by simulating natural selection, inheritance and variation in biological evolution. In multi-objective optimization, each individual in the population represents a set of potential solutions. The genetic algorithm will guide the search process by evaluating the fitness of each individual, selecting individuals with high fitness as parents to produce the next generation while maintaining the diversity of the population. This iteration continues until the optimal solution that meets all objectives and constraints is found.

[0115] According to the technical solution of the above embodiment of the present application, the technical solution of the present application solves the key problems in the energy management of extended-range vehicles through a double-layer structure, advanced data processing technology (GCN and LSTM), and advanced optimization algorithms (direct point matching method and genetic algorithm), and optimizes the battery life economy and fuel economy. The advantage of this method is that it can adapt to rapidly changing driving conditions while ensuring the efficiency and economy of energy use.

[0116] The following is an embodiment of the device of the present application, which can be used to execute the embodiment of the method of the present application. For details not disclosed in the embodiment of the device of the present application, please refer to the embodiment of the method of the present application.

[0117] Figure 7 Schematic diagram of the structure of the energy management device for extended-range vehicles provided in the embodiment of the present application. Figure 7 As shown, the range-extended vehicle energy management device includes:

[0118] The processing module 701 is configured to obtain historical vehicle speed information of the vehicle, and process the historical vehicle speed information using a pre-built vehicle speed prediction model to obtain a future predicted vehicle speed, wherein the vehicle speed prediction model processes spatial feature interactions through a graph convolutional network and processes time series prediction through a long short-term memory network;

[0119] The planning module 702 is configured to calculate the vehicle power demand and plan the battery state of charge change trajectory according to the vehicle power demand and the expected driving distance to obtain a battery state of charge reference trajectory;

[0120] The solution module 703 is configured to convert the preset multi-objective optimization control objective function into a nonlinear programming problem based on the future predicted vehicle speed and the battery state of charge reference trajectory, and use the genetic algorithm integrated with the elite strategy to solve the nonlinear programming problem to obtain the optimal energy management strategy, and use the optimal energy management strategy to manage the energy of the extended-range vehicle.

[0121] In some embodiments, Figure 7 Before using the pre-built vehicle speed prediction model to process the historical vehicle speed information, the construction module 704 obtains the historical vehicle speed vector and the historical acceleration vector, and divides the time window into multiple time segments, and uses the historical vehicle speed vector or historical acceleration vector corresponding to each time segment as a node, and the relationship between the nodes as an edge, and uses the nodes and the edges between the nodes to construct a graph structure, wherein the graph structure includes a vehicle speed graph structure and an acceleration graph structure.

[0122] In some embodiments, Figure 7The processing module 701 runs a graph convolutional network on the graph structure, uses the edges in the graph structure to transmit information, and uses the graph convolutional network to interactively learn the features between different nodes through the connection relationship between the nodes, so as to update the features of each node, wherein the updated node integrates the information of the adjacent nodes; the updated node is used as the input of the long short-term memory network, so that the long short-term memory network predicts the vehicle speed within a preset time period in the future according to the information contained in the updated node using a rolling window method.

[0123] In some embodiments, Figure 7 The processing module 701 updates the features of each node based on the information propagation mechanism in the graph convolutional network according to the edges defined in the graph structure; in the process of information propagation, the preset hyperparameters are used to adjust the interaction ratio between the vehicle speed feature and the acceleration feature to balance the contribution of feature update; information selection operations are performed, by weighted accumulation of node features updated through different information propagation steps, and the parameter matrix is ​​used to filter the information propagated in each layer, and the final filtered features are used as the input of the long short-term memory network.

[0124] In some embodiments, Figure 7 The planning module 702 calculates the vehicle power demand using a predetermined simplified vehicle power demand model to obtain the total power required by the vehicle, and associates the total power required by the vehicle with the driving distance so that the total power required by the vehicle is linearly correlated with the driving distance, and establishes a change trajectory of the battery state of charge based on the linear relationship between the total power required by the vehicle and the driving distance; wherein the simplified vehicle power demand model calculates the total power required by the vehicle based on the slope resistance, rolling resistance, air resistance and inertial resistance of the vehicle.

[0125] In some embodiments, Figure 7 The solution module 703 takes the battery state of charge as the state variable, takes the range extender output current as the control variable, sets constraints on the state variables and the control variables, and constructs the objective function of the multi-objective optimization control according to the state variables, the control variables and the constraints; and uses the direct point matching method to convert the objective function of the multi-objective optimization control into a nonlinear programming problem, wherein the direct point matching method includes using a piecewise polynomial equation to approximate the motion trajectory of the state variable so as to discretize the objective function of the multi-objective optimization control.

[0126] In some embodiments, Figure 7The solution module 703 initializes the population and generates a series of initial solutions, wherein each initial solution consists of control variables and state variables to be optimized; calculates individual fitness values, and applies a fast non-dominated sorting algorithm to stratify individuals in the population according to dominance relationships, and determines the dominance level and crowding degree of each individual; combines parent and child individuals, and selects individuals to form a new population according to the dominance level and crowding degree; generates new individuals through selection, crossover and mutation operations, repeats the generation operations of new populations and individuals, and finally selects the individual with the highest fitness as the optimal solution, which includes the optimal energy management strategy under the predetermined constraints.

[0127] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0128] Figure 8 Schematic diagram of the structure of the electronic device 8 provided in the embodiment of the present application. Figure 8 As shown, the electronic device 8 of this embodiment includes: a processor 801, a memory 802, and a computer program 803 stored in the memory 802 and executable on the processor 801. When the processor 801 executes the computer program 803, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor 801 executes the computer program 803, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0129] Exemplarily, the computer program 803 may be divided into one or more modules / units, which are stored in the memory 802 and executed by the processor 801 to complete the present application. The one or more modules / units may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program 803 in the electronic device 8.

[0130] The electronic device 8 may be a desktop computer, a notebook, a PDA, a cloud server, or other electronic device. The electronic device 8 may include, but is not limited to, a processor 801 and a memory 802. Those skilled in the art will appreciate that Figure 8 It is only an example of the electronic device 8 and does not constitute a limitation of the electronic device 8. It may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the electronic device may also include input and output devices, network access devices, buses, etc.

[0131] The processor 801 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or the processor may be any conventional processor, etc.

[0132] The memory 802 may be an internal storage unit of the electronic device 8, for example, a hard disk or memory of the electronic device 8. The memory 802 may also be an external storage device of the electronic device 8, for example, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 8. Further, the memory 802 may also include both an internal storage unit of the electronic device 8 and an external storage device. The memory 802 is used to store computer programs and other programs and data required by the electronic device. The memory 802 may also be used to temporarily store data that has been output or is to be output.

[0133] The technicians in the relevant field can clearly understand that for the convenience and simplicity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In practical applications, the above-mentioned function allocation can be completed by different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated in a processing unit, or each unit can exist physically separately, or two or more units can be integrated in one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, which will not be repeated here.

[0134] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0135] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0136] In the embodiments provided in the present application, it should be understood that the disclosed devices / computer equipment and methods can be implemented in other ways. For example, the device / computer equipment embodiments described above are only schematic. For example, the division of modules or units is only a logical function division. There may be other division methods in actual implementation. Multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0137] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0138] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0139] If the integrated module / unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium, and the computer program can implement the steps of the above-mentioned various method embodiments when executed by the processor. The computer program may include computer program code, which may be in source code form, object code form, executable file or some intermediate form. Computer-readable media may include: any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0140] The above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included in the protection scope of the present application.

Claims

1. A method for energy management of an extended-range vehicle, characterized in that: include: Acquire historical speed information of the vehicle, and process the historical speed information using a pre-built speed prediction model to obtain a future predicted speed, wherein the speed prediction model processes spatial feature interactions through a graph convolutional network to capture the spatial relationship between speed data, and processes time series prediction through a long short-term memory network; Calculate the vehicle power demand, and plan the battery state of charge change trajectory based on the vehicle power demand and the expected driving distance to obtain the battery state of charge reference trajectory; According to the future predicted vehicle speed and the battery state of charge reference trajectory, the preset multi-objective optimization control objective function is converted into a nonlinear programming problem, and the nonlinear programming problem is solved using a genetic algorithm that integrates an elite strategy to obtain an optimal energy management strategy, which is used to perform energy management on the extended-range vehicle.

2. The method according to claim 1, characterized in that Before processing the historical vehicle speed information using the pre-built vehicle speed prediction model, the method further includes: Obtain a historical vehicle speed vector and a historical acceleration vector, and divide the time window into multiple time segments, use the historical vehicle speed vector or historical acceleration vector corresponding to each time segment as a node, use the relationship between the nodes as an edge, and use the nodes and the edges between the nodes to construct a graph structure, wherein the graph structure includes a vehicle speed graph structure and an acceleration graph structure.

3. The method according to claim 2, characterized in that The method of processing the historical vehicle speed information using the pre-built vehicle speed prediction model to obtain the future predicted vehicle speed includes: Running the graph convolution network on the graph structure, using the edges in the graph structure to transfer information, and using the graph convolution network to interactively learn features between different nodes through the connection relationship between the nodes, so as to update the features of each node, wherein the updated node incorporates information of adjacent nodes; The updated nodes are used as inputs of the long short-term memory network, so that the long short-term memory network predicts the vehicle speed within a future preset time period using a rolling window method according to the information contained in the updated nodes.

4. The method according to claim 3, characterized in that The utilizing the graph convolutional network to interactively learn features between different nodes through the connection relationship between the nodes so as to update the features of each of the nodes includes: Based on the information propagation mechanism in the graph convolutional network, the features of each node are updated according to the edges defined in the graph structure; During the information propagation process, the interaction ratio between the vehicle speed feature and the acceleration feature is adjusted using the preset hyperparameters to balance the contribution of feature updates; An information selection operation is performed, by weighted accumulation of node features updated through different information propagation steps, and the information propagated in each layer is screened using a parameter matrix, and the final screened features are used as inputs of the long short-term memory network.

5. The method according to claim 1, characterized in that The calculation of the vehicle power demand and planning of the battery state of charge change trajectory according to the vehicle power demand and the expected driving distance include: The vehicle power demand is calculated using a predetermined simplified vehicle power demand model to obtain the total power required by the vehicle, and the total power required by the vehicle is associated with the driving distance so that the total power required by the vehicle is linearly correlated with the driving distance. According to the linear relationship between the total power required by the vehicle and the driving distance, a change trajectory of the battery state of charge is established; wherein the simplified vehicle power demand model calculates the total power required by the vehicle based on the slope resistance, rolling resistance, air resistance and inertial resistance of the vehicle.

6. The method according to claim 1, characterized in that The objective function of the preset multi-objective optimization control is converted into a nonlinear programming problem according to the future predicted vehicle speed and the battery state of charge reference trajectory, including: Taking the battery state of charge as a state variable and the range extender output current as a control variable, setting constraints on the state variables and the control variables, and constructing an objective function of the multi-objective optimization control according to the state variables, the control variables and the constraints; The objective function of the multi-objective optimization control is converted into a nonlinear programming problem using a direct point matching method, wherein the direct point matching method includes using a piecewise polynomial equation to approximate the motion trajectory of the state variable so as to discretize the objective function of the multi-objective optimization control.

7. The method according to claim 6, characterized in that The method of solving the nonlinear programming problem by using a genetic algorithm integrated with an elite strategy to obtain an optimal energy management strategy includes: Initialize the population and generate a series of initial solutions, where each initial solution consists of the control variables and state variables to be optimized; Calculate individual fitness values ​​and apply the fast non-dominated sorting algorithm to stratify individuals in the population according to dominance relationships and determine the dominance level and crowding degree of each individual; Combining the parent and offspring individuals, selecting individuals according to the dominance level and crowding degree to form a new population; New individuals are generated through selection, crossover and mutation operations, and the generation operations of new populations and individuals are repeated until the preset iteration termination conditions are reached. Finally, the individual with the highest fitness is selected as the optimal solution, which includes the optimal energy management strategy under the predetermined constraints.

8. An energy management device for an extended-range vehicle, characterized in that: include: a processing module configured to obtain historical vehicle speed information of the vehicle, and process the historical vehicle speed information using a pre-built vehicle speed prediction model to obtain a future predicted vehicle speed, wherein the vehicle speed prediction model processes spatial feature interactions through a graph convolutional network to capture the spatial relationship between vehicle speed data, and processes time series prediction through a long short-term memory network; A planning module is configured to calculate the power demand of the vehicle and plan the change trajectory of the battery state of charge according to the power demand of the vehicle and the expected driving distance to obtain a reference trajectory of the battery state of charge; The solution module is configured to convert the preset multi-objective optimization control objective function into a nonlinear programming problem according to the future predicted vehicle speed and the battery state of charge reference trajectory, and solve the nonlinear programming problem using a genetic algorithm that integrates an elite strategy to obtain an optimal energy management strategy, and use the optimal energy management strategy to perform energy management on the extended-range vehicle.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images