A hydrogen fuel cell vehicle real-time energy management architecture based on vehicle-to-road information

Abstract

This embodiment discloses a real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information. It receives multi-dimensional time-series data, including vehicle operating status data and traffic status data, from a cloud-based traffic control platform. A Long Short-Term Memory (LSTM) network is used to obtain the predicted speed of the target hydrogen fuel cell vehicle. A multi-task LSTM prediction model is employed to obtain the predicted instantaneous hydrogen consumption rate, predicted fuel cell health status, and predicted operating efficiency of the hydrogen fuel cell, thereby obtaining the optimal trajectory sequence for vehicle speed and the optimal output power command sequence for the fuel cell. Furthermore, power allocation commands between the hydrogen fuel cell and the power battery are obtained and sent to the underlying architecture for execution by onboard execution devices. Based on this management architecture, the hydrogen fuel cell system can respond rapidly, improving the lifespan of the hydrogen fuel cell and hydrogen energy efficiency, making it particularly suitable for hydrogen fuel cell vehicles operating under long-distance transportation conditions.

Description

Technical Field

[0001] This invention relates to the field of hydrogen fuel cell vehicle technology, and in particular to a real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information. Background Technology

[0002] As new energy vehicles continue to be used in the transportation sector, the lag and uncertainty in infrastructure construction faced by transportation systems are becoming increasingly prominent, and traditional transportation system coordination and vehicle energy control technologies are gradually showing significant shortcomings. In September 2025, eight departments, including the Ministry of Industry and Information Technology, issued the "Work Plan for Stabilizing Growth in the Automobile Industry (2025-2026)," proposing the construction of hydrogen energy infrastructure and the industrial application of intelligent connected vehicle technology. This plan aims to actively promote the demonstration application of fuel cell vehicles, drive the large-scale development of medium- and long-distance fuel cell commercial vehicles, conduct in-depth pilot applications of "vehicle-road-cloud integration" for intelligent connected vehicles, accelerate the construction of connected infrastructure and cloud control platforms, and encourage the pre-installation of high-performance communication modules such as V2X and 5G in automobiles.

[0003] Hydrogen fuel cell vehicles (FCEVs) use hydrogen fuel cells as a crucial device for hydrogen-to-electricity conversion. Hydrogen refueling requires specially designated infrastructure. As a clean energy source, hydrogen produces only a small amount of industrial waste gas during its conversion to electricity, effectively improving the environmental friendliness of transportation vehicles. Following the release of the "Medium- and Long-Term Plan for the Development of the Hydrogen Energy Industry (2021-2035)" in March 2022, hydrogen fuel cell vehicles entered commercial operation. Their power comes from hydrogen, which has multiple acquisition methods and high energy density, and they can utilize onboard devices to connect to the "vehicle-road-cloud" system to obtain more comprehensive real-time traffic data.

[0004] However, compared to traditional gasoline vehicles, hydrogen fuel cell vehicles, despite their high energy density and environmental friendliness, exhibit a relatively slow dynamic response in their hydrogen fuel cell systems. Current energy management strategies primarily rely on the vehicle's current state to respond to impending road condition changes, but frequent load adjustments reduce battery life and hydrogen efficiency. Furthermore, current vehicle-to-everything (V2X) technologies are mostly applied to driver assistance systems, lacking in-depth application in optimizing and managing vehicle powertrain systems. Summary of the Invention

[0005] This invention discloses a real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information, in order to overcome the aforementioned technical problems.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: A real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information includes a vehicle energy management system; the vehicle energy management system includes a top-level architecture, a middle-level architecture, and a bottom-level architecture. The underlying architecture is a hydrogen fuel cell vehicle; the hydrogen fuel cell vehicle includes an on-board controller and on-board actuators; the on-board controller is used to monitor and store vehicle operating status data in real time; The intermediate layer architecture consists of roadside units and traffic facilities; the roadside units and traffic facilities are used to acquire traffic status data and receive dispatch signals from the cloud-based traffic control platform. The top-level architecture is a cloud-based traffic control platform; the cloud-based traffic control platform includes a cloud data center and a cloud server; the cloud data center stores map data; the cloud server is used to obtain the predicted speed of the target hydrogen fuel cell vehicle using a long short-term memory network based on map data, vehicle operating status data, and traffic status data, and to obtain the predicted instantaneous hydrogen consumption rate, predicted fuel cell health status, and predicted operating efficiency of the hydrogen fuel cell using a multi-task LSTM prediction model, so as to obtain the optimal trajectory sequence of vehicle speed and the optimal output power command sequence of the fuel cell, and then obtain the power allocation command between the hydrogen fuel cell and the power battery, and send it to the hydrogen fuel cell vehicle for execution by the on-board execution equipment.

[0007] Furthermore, the management method for the real-time energy management architecture of hydrogen fuel cell vehicles based on vehicle-road information includes the following steps: S1: Obtain vehicle operating status data of the target hydrogen fuel cell vehicle through the on-board controller of the target hydrogen fuel cell vehicle; S2: Using a cloud-based traffic control platform and a roadside unit and traffic facilities with an intermediate layer architecture, environmental event sequence data is acquired to obtain time series data with multidimensional characteristics of the target hydrogen fuel cell vehicle based on the vehicle operation status data of the target hydrogen fuel cell vehicle. S3: Based on the vehicle operating status data and environmental event sequence data of the target hydrogen fuel cell vehicle, a long short-term memory network is used to obtain the LSTM hidden state sequence, so as to obtain the vehicle driving condition time sequence in the future main prediction time domain, and then obtain the predicted speed of the target hydrogen fuel cell vehicle. S4: Based on the LSTM hidden state sequence, a multi-task LSTM prediction model is used to obtain the predicted instantaneous hydrogen consumption rate, the predicted health status of the fuel cell, and the predicted operating efficiency of the hydrogen fuel cell. S5: Based on the predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell, the predicted health status of the fuel cell, and the predicted operating efficiency, construct an objective function with the total cost as the optimization objective. Then, use model predictive control to obtain the optimal trajectory sequence of vehicle speed that minimizes the objective function and the optimal output power command sequence of the fuel cell. S6: Establish an objective function and vehicle dynamics model that minimizes the output power of the hydrogen fuel cell. Based on the optimal trajectory sequence of vehicle speed and the optimal output power command sequence of the hydrogen fuel cell, obtain the power allocation command between the hydrogen fuel cell and the power battery, i.e. the output power of the hydrogen fuel cell and the output power of the power battery, and send it to the underlying architecture for execution by the on-board execution device.

[0008] Furthermore, the vehicle operating status data of the target hydrogen fuel cell vehicle includes: standardized vehicle speed, hydrogen storage capacity of the hydrogen storage tank, and state of charge of the power battery; and boundary constraints are set for the vehicle operating status data. The environmental event sequence data includes: the slope, curvature, traffic light status set, current charging price, and charging waiting time on the road along the target hydrogen fuel cell vehicle's driving trajectory.

[0009] Furthermore, S3 includes: a main state coding branch and a secondary state coding branch; S31: The main state encoding branch of the LSTM dual-branch structure prediction network is used to obtain the hidden state of the vehicle operating state sequence based on the vehicle operating state data of the target hydrogen fuel cell vehicle. S32: The sub-state encoding branch of the prediction network using the LSTM dual-branch structure is used to obtain the hidden state of the environmental event sequence based on the environmental event sequence data; S33: Obtain the merged LSTM hidden state sequence based on the hidden states of the vehicle operation state sequence and the hidden states of the environmental event sequence; S34: Based on the merged LSTM hidden state sequence, obtain the vehicle driving condition time series of the target hydrogen fuel cell vehicle in the future main prediction time domain, so as to obtain the predicted speed of the target hydrogen fuel cell vehicle.

[0010] Furthermore, the predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell is expressed as follows:

[0011] In the formula: The predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell; The mapping function is obtained through learning; This refers to the output power of the hydrogen fuel cell; Power required by the vehicle; SOC This refers to the state of charge of the power battery. The control cycle time for the operation of a hydrogen fuel cell; The predicted health status of the fuel cell is represented as follows:

[0012] In the formula: For the predicted health status of fuel cells; The control cycle time for the operation of a hydrogen fuel cell; N The total number of control cycles for the operation of the hydrogen fuel cell; For the first The current of the power battery within each control cycle; This is the rated current of the hydrogen fuel cell; For the first Reference temperature of the hydrogen fuel cell within each control cycle; For the first The actual temperature of the power battery within each control cycle; The predicted operating efficiency is expressed as follows: (13) In the formula: For the first k The predicted operating efficiency of the vehicle within each control cycle; , , All are weighting coefficients, satisfying ; , , These are the efficiency of the hydrogen fuel cell, the efficiency of the power battery, and the instantaneous efficiency of the motor.

[0013] Furthermore, the objective function for minimizing the total cost is expressed as follows:

[0014] In the formula: For at any time t The total cost optimization objective function; The primary prediction time domain; The control cycle time for the operation of a hydrogen fuel cell; This is the weighting coefficient for the hydrogen consumption cost tracking error; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; The weighting coefficient for the vehicle operating efficiency tracking error; For hydrogen fuel cells in the first k The output power within each control cycle is Instantaneous hydrogen consumption rate at time; For the first Equivalent factor for each control cycle; For the predicted health status of fuel cells; For the first The predicted operating efficiency of the vehicle within each control cycle; The weighting coefficient for the target speed of the vehicle; For at any time Predicted vehicle speed; The target speed for the vehicle; in,

[0015] In the formula: For the first k Equivalent factor for each control cycle; The coefficient is positively correlated with the degree of fluctuation in the predicted vehicle speed; This serves as a reference value for the health status of hydrogen fuel cells. This is a transpose.

[0016] Furthermore, the output power of the hydrogen fuel cell The minimum objective function is expressed as follows:

[0017] In the formula: This is the weighting coefficient for the hydrogen consumption cost tracking error; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; The weighting coefficient for the vehicle operating efficiency tracking error; For time step; The weighting coefficient for the target speed of the vehicle; The primary prediction time domain; For the predicted health status of hydrogen fuel cells; For the first The predicted operating efficiency of the vehicle within each control cycle; The weighting coefficient for the target speed of the vehicle; For at any time Predicted vehicle speed; The target speed for the vehicle; This represents the predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell.

[0018] Furthermore, the vehicle dynamics model is represented as follows:

[0019] In the formula: air density; Road slope; For the efficiency of the transmission system; The derivative of the vehicle speed, i.e. ; For the first k The differential value of the vehicle speed within each control cycle, i.e., the instantaneous acceleration within that cycle. ; Indicates the first kThe actual output power of the hydrogen fuel cell within each control cycle.

[0020] Furthermore, the weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation is dynamically adjusted based on the current or temperature of the hydrogen fuel cell. When the current or temperature of the hydrogen fuel cell exceeds a set threshold,

[0021] in, Indicates the safety threshold; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; This represents the initial value for the hydrogen fuel cell lifespan weighting coefficient. The safety threshold for the health status of hydrogen fuel cells; This represents the lower threshold for the health status of a hydrogen fuel cell.

[0022] Beneficial Effects: This invention provides a real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information. It receives multi-dimensional time-series data, including vehicle operating status data and traffic status data, from a cloud-based traffic control platform. A Long Short-Term Memory (LSTM) network is used to obtain the predicted speed of the target hydrogen fuel cell vehicle. A multi-task LSTM prediction model is employed to obtain the predicted instantaneous hydrogen consumption rate, predicted fuel cell health status, and predicted operating efficiency of the hydrogen fuel cell. This yields the optimal trajectory sequence for vehicle speed and the optimal output power command sequence for the fuel cell. Furthermore, power allocation commands between the hydrogen fuel cell and the power battery are obtained and sent to the underlying architecture for execution by onboard devices. Based on this management architecture, the hydrogen fuel cell system can respond rapidly, improving fuel cell lifespan and hydrogen energy efficiency, making it particularly suitable for hydrogen fuel cell vehicles operating under long-distance transportation conditions. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0024] Figure 1 This is a schematic diagram of the overall architecture of the hydrogen fuel cell vehicle energy management system based on vehicle-road cooperative information of the present invention; Figure 2 This is a flowchart illustrating the energy management strategy for hydrogen fuel cell vehicles under long-distance transportation conditions in an embodiment of the present invention. Figure 3 Set parameters for the vehicle controller in this embodiment of the invention; Figure 4 This is a schematic diagram of the processing flow of time series data with multidimensional characteristics of the target hydrogen fuel cell vehicle in an embodiment of the present invention. Figure 5 This is a schematic diagram of the vehicle dynamics model in an embodiment of the present invention; Figure 6 This is a framework diagram of the vehicle-road-cloud service simulation platform in an embodiment of the present invention; Figure 7 This is a flowchart illustrating the network performance testing of the simulation platform in this embodiment of the invention. Figure 8 This is a hardware architecture diagram of the real vehicle testing platform in an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0026] This embodiment introduces a real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information, including a vehicle energy management system; the vehicle energy management system includes a top-level architecture, a middle-level architecture, and a bottom-level architecture. The underlying architecture is for hydrogen fuel cell vehicles; the hydrogen fuel cell vehicles include onboard controllers and onboard actuators. The on-board controller is used to monitor and store vehicle operating status data in real time. The vehicle operating status data includes vehicle parameters, hydrogen fuel cell parameters, and power battery system parameters. The hydrogen fuel cell system parameters include the output power and state of charge of the hydrogen fuel cell; the power battery system parameters include the output power and state of charge of the lithium battery, hydrogen storage tank pressure, vehicle speed, and other vehicle operating information. The hydrogen fuel cell vehicle of this embodiment includes an onboard controller and onboard actuators. The onboard actuators include a hydrogen storage tank, a hydrogen fuel cell, a power battery, a DC / DC converter, a DC / AC converter, and a motor. The onboard controller monitors and stores vehicle operating status data in real time, including the output power and state of charge of the hydrogen fuel cell / lithium battery, hydrogen storage tank pressure, and vehicle speed. The onboard controller connects to a 4G / 5G communication network via a built-in vehicle-to-infrastructure (V2I) communication interface, uploading vehicle operating status data to a cloud-based traffic control platform and receiving control signals from the platform. This allows for coordinated control of the onboard actuators, enabling the vehicle to achieve optimal performance in key indicators such as hydrogen consumption, battery life, and operating efficiency while meeting driving requirements.

[0027] The middle layer architecture consists of roadside units and traffic facilities; the roadside units and traffic facilities are used to acquire traffic status data; the traffic status data includes road and intersection information, availability status of hydrogen refueling / charging stations, energy prices, traffic light status information, and traffic warning information from roadside electronic displays; as well as receiving dispatch signals from the cloud-based traffic control platform; Specifically, the middle layer consists of roadside units (RSUs) and traffic facilities, including hydrogen refueling / charging stations, traffic lights, roadside electronic displays, etc. It communicates with the cloud-based traffic control platform in real time and provides feedback on road and intersection information, the availability status of hydrogen refueling / charging stations, energy prices, traffic light status information, and traffic warning information from the roadside electronic displays. At the same time, it receives dispatch signals from the cloud-based traffic control platform.

[0028] The top-level architecture is a cloud-based traffic control platform; the cloud-based traffic control platform includes a cloud data center and a cloud server; the cloud data center stores map data; the cloud server is used to obtain control signals based on map data, vehicle operating status data, and traffic status data and send them to the underlying architecture and on-board execution devices; Specifically, the cloud-based traffic control platform in this embodiment includes a cloud data center and a cloud server. The cloud data center stores historical and real-time traffic information such as map data, cluster vehicle operation status data, and road information. The cloud server receives real-time feedback information from the middle layer architecture and the bottom layer architecture. After performing predictive processing and optimization decisions on traffic status data, hydrogen refueling / charging station data, and vehicle operation status data, it sends control signals to the bottom layer.

[0029] The vehicle energy management system in this embodiment mainly adopts a three-layer collaborative architecture of "cloud-road-vehicle", such as Figure 1As shown, vehicle energy optimization management is achieved through multiple types of energy and information flows. The top-level architecture is a cloud-based traffic control platform, the middle-level architecture consists of roadside units (RSUs) and traffic facilities, and the bottom-level architecture is the hydrogen fuel cell vehicle. Short dashed lines and solid lines represent information flows, thick solid lines represent mechanical energy, long dashed lines represent gas energy, and thin solid lines represent electrical energy. This multi-faceted energy and information flow enables optimized vehicle energy management. This embodiment also discloses a management method for a real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information, applicable to long-distance transportation conditions, such as... Figure 2 As shown, it includes the following steps: S1: After the driver starts the vehicle, the vehicle's operating status data, including the standardized vehicle speed, is obtained through the onboard controller of the target hydrogen fuel cell vehicle. Hydrogen storage capacity of hydrogen storage tank and the state of charge of the power battery And set boundary constraints for vehicle operating status data.

[0030] Specifically, the hydrogen fuel cell vehicle energy management system is initialized, parsing vehicle and top-level information and configuring driving parameters, including reading inherent vehicle parameters, initializing the vehicle-to-infrastructure (V2I) communication interface, and configuring controller parameters. In this embodiment, vehicle parameters include the vehicle's curb weight. m Wind resistance parameters C d Rated current of hydrogen fuel cells windward area A Rolling resistance coefficient f、 Vehicle speed, etc. Hydrogen fuel cell system parameters include hydrogen storage tank pressure. p H Minimum operating power of hydrogen fuel cells P fc.min Peak power P fc.max Hydrogen consumption rate Δm H、 Hydrogen storage capacity of hydrogen storage tank Etc. Power battery system parameters include the total capacity of the power battery. Q bat Nominal voltage U The upper and lower limits of the state of charge (SOC) of the power battery and After reading the inherent parameters of the hydrogen fuel cell vehicle, boundary constraints are set for the vehicle operating status data: (1) (2) (3) In the formula: This represents the minimum operating power for a hydrogen fuel cell; For the first The actual output power of the hydrogen fuel cell within each control cycle; This represents the peak power of the hydrogen fuel cell; This refers to the minimum permissible discharge power of the power battery. For the first The actual charging and discharging power of the power battery within each control cycle; This refers to the maximum permissible discharge power of the power battery. This is the lower limit threshold for the state of charge of the power battery; For the first Real-time state of charge of the power battery within each control cycle; This is the upper limit threshold for the state of charge of the power battery; The control cycle time for the operation of a hydrogen fuel cell; Specifically, the vehicle-to-infrastructure (V2I) communication interface of the onboard controller is initialized, the V2I communication module is started, and communication is established with the cloud-based traffic control platform. It receives map data, real-time traffic information such as roads, and control signals from the top-level architecture, parses and verifies the received data, and extracts traffic constraints from the planned roads. (Reference) Figure 3 Configure the parameters of the on-board controller, including the weights of the optimization target and the parameters of predictive control.

[0031] S2: Employ a cloud-based traffic control platform to acquire environmental event sequence data, and based on the vehicle operating status data of the target hydrogen fuel cell vehicle, acquire time series data of the target hydrogen fuel cell vehicle with multidimensional characteristics. The environmental events include the slope of the road along the target hydrogen fuel cell vehicle's driving trajectory. curvature Traffic signal light status set Current charging price Charging wait time ; The cloud-based traffic control platform in this embodiment processes map data from the cloud data center, traffic status data from the intermediate layer, and vehicle operation status data from the underlying vehicle controller, and generates vehicle-road-cloud services suitable for hydrogen fuel cell vehicles.

[0032] Specifically, such as Figure 4The cloud server collects data from different data sources, including data from the cloud data center, the middleware framework (environmental event sequence data), and the underlying framework (vehicle controller data). All this information is mapped to the same spatiotemporal framework. Based on a time reference system and according to the different sampling intervals of each type of source data, spline interpolation is used to establish a unified time period corresponding to different data sources. Δt A continuous sequence of data; based on a spatial reference frame, establish the correspondence between the complete set of data information with obvious coordinates in the spatial dimension and the same spatial reference frame.

[0033] Specifically, the collected road and intersection information includes: No. Traffic light status dataset at each intersection , means as follows: (4) In the formula: For the first n Dataset of traffic light status at an intersection; t Represents a global time variable; , The first n At the intersection t The color of the traffic light at a given moment and its remaining duration are a time-varying factor. t Changing dynamic quantity; For indexing road intersections; The monitoring equipment, such as cameras and lidar located at road intersections on the traffic map, collects information in real time, including vehicle tag numbers, driving trajectories, and speeds. This information is used to describe road intersections and roads on the traffic map as nodes and edges, respectively. Represented as the first n At the intersection of the road, the first road in the area n The first one appeared at the intersection of the road. m One vehicle t Datasets uploaded in real time Described as: (5) In the formula: Indicates the first n The first one appeared at the intersection of the road. One vehicle t Data sets that are uploaded in real time; An identifier representing the observed vehicle; Indicates the coordinate position of the observed vehicle; This indicates the speed of the observed vehicle, expressed in m / s. n For indexing road intersections; For the first nAn index of vehicles appearing at each road intersection; Specifically, regarding time t The j Data set of availability status and energy prices for individual hydrogen refueling / charging stations. Described as: (6) In the formula: Indicates time t The j A dataset of the availability status and energy prices of several hydrogen refueling / charging stations; Identifier for hydrogen refueling / charging stations; Indicates the coordinates of the energy station; Indicates the energy type, either hydrogen or electricity; Indicates the current status, with the type being Idle / Busy / Maintenance; This indicates the current charging price, expressed in yuan / kg or yuan / kWh. This indicates the charging wait time, in minutes. j Index for hydrogen refueling / charging stations; Specifically, the vehicle operation data is presented as follows: (7) In the formula: For the first m A dataset of vehicle operation data; and The first m The coordinates and speed of each vehicle; For the first m Hydrogen storage capacity of each vehicle; m For the first n An index of vehicles appearing at each road intersection; Specifically, cloud servers map multi-source data received from cloud data centers, middleware architecture, and underlying architecture to a unified time and space reference system. Firstly, in terms of the time reference system, the timestamps of the multi-source data need to be... t Unified as a network time protocol for the cloud platform T p That is, the main prediction time domain, and considering the differences in data collection frequency from different data sources, spline interpolation is used to generate the same time period. A continuous data sequence. Taking vehicle speed as an example, within the interval... Constructing a cubic polynomial To satisfy , Then, in the spatial reference frame, all data with spatial attributes, including the coordinates of the observed vehicle, are... Coordinates of the energy station Vehicle location coordinates on the traffic map This is mapped onto the physical map, forming a unified spatial reference coordinate system. For the first k Each sampling time; For controlling the periodic index; Specifically, based on historical data from the cloud data center, the cloud server removes and marks data in the vehicle status that exceeds a threshold, and then uses linear interpolation to fill in the missing data points.

[0034] Specifically, in order to remove the influence of multi-source data with different features on the subsequent algorithm prediction performance, the cloud server performs standardization processing on multi-source data with numerical distributions of Gaussian distribution, interval distribution, and categorical features: For a data sequence of vehicle speeds with a Gaussian distribution, Z-score standardization is performed, and the standardized values ​​are... Described as: (8) In the formula: Standardized value for vehicle speed; This is the raw data for vehicle speed; and represents the mean and standard deviation of the vehicle speed feature on the dataset, respectively.

[0035] For data sequences containing hydrogen storage capacity of vehicle hydrogen storage tanks and state of charge of power batteries, whose numerical distributions exhibit interval distribution characteristics, the minimum value of the data is determined. and maximum value Perform 0-1 standardization, and its standardized value Described as: (9) In the formula: These are standardized values ​​for hydrogen storage capacity and state of charge; This provides raw data on hydrogen storage capacity and state of charge. This represents the minimum values ​​of hydrogen storage capacity and state of charge characteristics on the dataset. This represents the maximum value of hydrogen storage capacity and state of charge characteristics on the dataset; Data on the energy station's charging type, current status, and traffic light color and direction of reference are converted into binary form.

[0036] S3: Based on the multi-dimensional time series data of the target hydrogen fuel cell vehicle, a Long Short-Term Memory (LSTM) network is used to predict the vehicle's driving status and obtain the LSTM hidden state sequence to obtain the vehicle's driving condition time series in the future main prediction time domain Tp, so as to obtain the predicted speed of the target hydrogen fuel cell vehicle. Specifically, cloud servers are used and LSTM is used to predict the future driving status of vehicles. The optimization objectives are to minimize hydrogen consumption, slow down battery life degradation, and maximize vehicle operating efficiency. The results of the calculation of the vehicle's global optimal driving trajectory and energy management parameters are obtained.

[0037] This embodiment employs an LSTM dual-branch prediction network to obtain the main prediction time domain. T p Operating conditions of the target hydrogen fuel cell vehicle; Specifically, by using multi-dimensional time series historical data, the target hydrogen fuel cell vehicle is predicted for the next main prediction time domain. T p The driving conditions within the vehicle. For the target vehicle, the time series data of the target hydrogen fuel cell vehicle with multidimensional features collected at present is used as the input of the LSTM network. The time series data of the target hydrogen fuel cell vehicle with multidimensional features is composed of the following three types of features in the historical time window. The specific features include: (1) the vehicle operating status of the target hydrogen fuel cell vehicle: the standardized vehicle speed. Hydrogen storage capacity of hydrogen storage tank and the state of charge of the power battery (2) Road characteristics on the planned route: Based on the planned route data issued by the cloud-based traffic control platform, the slope of the road on the vehicle trajectory of the target hydrogen fuel cell vehicle is obtained. and curvature (3) Dynamic traffic event information: a set of traffic light statuses obtained through the vehicle-road cooperative communication interface. (4) Energy site characteristics: current charging price Charging wait time .

[0038] Preferably, S3 includes: a main state coding branch and a secondary state coding branch; S31: The main state encoding branch of the LSTM dual-branch prediction network is used to obtain the hidden state of the vehicle operating state sequence based on the vehicle operating state data of the target hydrogen fuel cell vehicle. ; S32: The sub-state encoding branch of the prediction network using an LSTM dual-branch structure is used to obtain the hidden states of the environmental event sequence based on the environmental event sequence data. ; S33: Hidden state based on vehicle operating state sequence Hidden states of environmental event sequences Obtain the merged LSTM hidden state sequence; S34: Obtain the future master prediction time domain of the target hydrogen fuel cell vehicle based on the merged LSTM hidden state sequence. T p The vehicle's driving conditions time series are used to obtain the predicted speed of the target hydrogen fuel cell vehicle. Among them, the future main prediction time domain of the target hydrogen fuel cell vehicle is obtained. T p The formula used for the vehicle driving condition time series is as follows: (10) In the formula: For the first v Vehicle in the future main prediction time domain T p Time series of vehicle driving conditions within the system; This is the weight matrix. It is the bias vector; This is the sequence of hidden states after merging with LSTM.

[0039] Among them, the predicted vehicle driving condition time series This includes predicted vehicle speed sequences and acceleration sequences, which serve as the basis for predicting future energy management.

[0040] In this embodiment, the LSTM dual-branch prediction network includes a main state-coding branch and a secondary state-coding branch; the main state-coding branch processes vehicle operating state data, i.e., the input is the vehicle operating state data of the target hydrogen fuel cell vehicle within a historical time window, i.e., continuous data. dimensional feature time series data, i.e. The hidden states of the vehicle's operating state sequence are extracted using an LSTM network. ;in, For the first v Time series of vehicle operating status characteristics; It is the set of real numbers; This refers to the length of the historical time window, i.e., the number of time steps. The number of vehicle operating state features, i.e., the dimension of vehicle operating state features; the sub-state coding branch processes environmental state data, including physical map features (road slope, curvature, traffic light status) and energy station features (charging price, charging waiting time), i.e., the input is a set of features. Dimensional feature environmental event sequence data Hidden states of environmental event sequences are extracted using an LSTM dual-branch prediction network. This involves merging the time series data from the two branches, i.e. ,in, For the first v Time series of environmental state characteristics of vehicles; The number of environmental state characteristics, i.e., the dimension of environmental state characteristics; The sequence of hidden states in the merged LSTM; The hidden state output by the main state encoding branch; The hidden state is encoded by the sub-state encoding branch; then the decoder will... Mapped to the future master prediction time domain T p The predicted vehicle driving condition time series.

[0041] S4: Based on the LSTM hidden state sequence, a multi-task LSTM prediction model is used to obtain the predicted instantaneous hydrogen consumption rate, the predicted health status of the fuel cell, and the predicted operating efficiency of the hydrogen fuel cell. Specifically, the hidden states extracted by the LSTM dual-branch structure prediction network... The input is fed into a multi-task LSTM prediction model, where the hidden state sequence is... The data includes vehicle operating status within a time window, road geometry features, dynamic traffic event information, and energy station characteristics. A multi-task LSTM prediction model collectively stores the LSTM hidden state sequence. As input, and with multiple sub-task networks set to predict the instantaneous hydrogen consumption rate of the hydrogen fuel cell, fuel cell life, and vehicle operating efficiency respectively, the following is to be implemented. The data were input into three sub-task networks respectively, resulting in: (1) Prediction of the instantaneous hydrogen consumption rate of the hydrogen fuel cell: (2) Predicted health status of fuel cells: (3) Predicted operating efficiency: .in, , , For the mapping function of the subtask network, , , These are the learnable parameters for the corresponding subnetwork.

[0042] Preferably, the predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell is expressed as follows: This embodiment outputs the future primary prediction time domain. T p Instantaneous hydrogen consumption rate Hydrogen consumption rate of hydrogen fuel cells was obtained by polynomial fitting. The relationship with other vehicle technical specifications is described as follows: (11) In the formula: The predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell; The mapping function is obtained through learning; This refers to the output power of the hydrogen fuel cell, measured in kW. The required power output for the vehicle is expressed in kW; the State of Charge (SOC) is the state of charge of the battery, expressed in %. The control cycle time for the operation of a hydrogen fuel cell; Preferably, the predicted health status of the fuel cell is expressed as follows: Specifically, the assessment target for fuel cell lifespan is health status. This predicted value is calculated based on parameters such as battery current and temperature, and is described as follows: (12) In the formula: To predict the health status of the fuel cell, The fuel cell lifespan gradually decreases from 100% until it falls below a certain set value. The control cycle time for the operation of a hydrogen fuel cell; N The total number of control cycles for the operation of the hydrogen fuel cell; For the first The current of the power battery within each control cycle; This is the rated current of the hydrogen fuel cell; For the first Reference temperature of the hydrogen fuel cell within each control cycle; For the first The actual temperature of the power battery within each control cycle; Preferably, the predicted operating efficiency is expressed as follows: Specifically, the predicted operating efficiency is a weighted combination of the efficiency of the hydrogen fuel cell, the efficiency of the power battery, and the instantaneous efficiency of the motor, which is predicted by the sub-task network based on the current state, thus determining the vehicle's predicted operating efficiency. for: (13) In the formula: For the first kThe predicted operating efficiency of the vehicle within each control cycle, with a value between 0 and 1; , , All are weighting coefficients, satisfying ; , , These are the efficiency of the hydrogen fuel cell, the efficiency of the power battery, and the instantaneous efficiency of the motor.

[0043] S5: Based on the predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell, the predicted health status of the fuel cell, and the predicted operating efficiency, construct an objective function with the total cost as the optimization objective. Then, use Model Predictive Control (MPC) to obtain the optimal trajectory sequence of vehicle speeds that minimizes the objective function. And the optimal output power command sequence (energy control command) for the fuel cell. .

[0044] Preferably, the objective function for minimizing the total cost is expressed as follows: (14) In the formula: For at any time t The total cost optimization objective function; The primary prediction time domain; The control cycle time for the operation of a hydrogen fuel cell; This is the weighting coefficient for the hydrogen consumption cost tracking error; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; The weighting coefficient for the vehicle operating efficiency tracking error; For hydrogen fuel cells in the first k The output power within each control cycle is Instantaneous hydrogen consumption rate at time; For the first Equivalent factor for each control cycle; For the predicted health status of fuel cells; For the first The predicted operating efficiency of the vehicle within each control cycle; The weighting coefficient for the target speed of the vehicle; For at any time Predicted vehicle speed; The target speed for the vehicle; The control variables are: (15) In the formula: For the first System control variables for each control cycle; For the first The actual output power of the hydrogen fuel cell within each control cycle; The equivalence factor is calculated using a bisection iterative method, as shown below: (16) In the formula: For the first k Equivalent factor for each control cycle; The coefficient is positively correlated with the degree of fluctuation in the predicted vehicle speed; This serves as a reference value for the health status of hydrogen fuel cells. For transpose; Specifically, at any given moment t First, utilize the data updated in the previous cycle. Solving the objective function yields the control command for the current cycle. After execution, the new state is observed, and then updated based on the SOH deviation. Used for the next cycle.

[0045] Specifically, the cloud server constructs an objective function based on the vehicle's current operating status and historical information from the cloud data center, considering boundary constraints on vehicle operating status data and safety constraints such as traffic lights and vehicle speed. The objective function aims to minimize total cost while also taking into account safety constraints (boundary constraints on vehicle operating status data). This objective function is used in the prediction time domain. Internally, optimize the output power of hydrogen fuel cells. The sequence of values ​​is used to minimize the weighted sum of hydrogen consumption, battery life degradation, operating efficiency loss, and vehicle speed tracking deviation. In the objective function, the first term is the sum of hydrogen consumption, battery life degradation, and operating efficiency for each control cycle within the prediction time domain, and the second term is the deviation between vehicle speed and target speed. Control variables... By influencing , and To change the objective function value, solve for the optimization result. The command for the first control cycle is sent to the vehicle controller.

[0046] In this embodiment, the system state vector is constructed as follows: (17) In the formula: For the first The system state vector for each control cycle; The predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell; For the predicted health status of fuel cells; For the first k The predicted operating efficiency of the vehicle within each control cycle; The control cycle time for the operation of a hydrogen fuel cell; For transpose; In this embodiment, the constructed objective function is combined with the equivalent factor updated in each control cycle. The optimization algorithm in model predictive control is used to solve for the solution. The optimal system control variables and their corresponding vehicle state trajectories are minimized, thereby obtaining the optimal trajectory sequence of vehicle speed output by the cloud server. And the optimal output power command sequence (energy control command) for the fuel cell. .

[0047] Among them, the optimal trajectory sequence of vehicle speed This refers to the vehicle's future prediction time domain. The time series consisting of the optimal speed values ​​at each moment within the vehicle's orbit is essentially a curve representing the vehicle's speed over time. The optimal vehicle speed trajectory is transmitted to the onboard controller via 4G / 5G communication and is described as follows: (18) In the formula: The optimal trajectory sequence for vehicle speed; For at any time t The target speed of the vehicle; the entire sequence constitutes a speed trajectory.

[0048] Energy control commands are used to instruct the onboard controller to allocate the output power of the hydrogen fuel cell, and are described as follows: (19) In the formula: This is the optimal output power command sequence for the hydrogen fuel cell; For at any time t The target output power of the fuel cell; S6: Optimal trajectory sequence based on vehicle speed And the optimal output power command sequence for hydrogen fuel cells. Obtain the output power of the execution unit; In this embodiment, the vehicle controller uses the optimal trajectory sequence of vehicle speed received from the cloud. And the optimal output power command sequence (energy control command) for hydrogen fuel cells. Based on the information and combined with the real-time vehicle status, energy is allocated through energy management strategies, and the output power of the execution unit is output. The specific technical solution is as follows: The vehicle controller receives the optimal trajectory sequence of vehicle speed from the cloud server in real time via the vehicle-to-infrastructure (V2I) communication interface. And the optimal output power command sequence for hydrogen fuel cells. Meanwhile, the vehicle controller collects vehicle status data in real time via the CAN bus. (20) In the formula: For at any time t The system state vector; The on-board controller performs energy allocation using a rolling time-domain optimization method and adopts an objective function that minimizes the output power of the hydrogen fuel cell. It obtains a vehicle dynamics model using inherent vehicle parameters and road information, and solves for the power allocation commands of the hydrogen fuel cell and the power battery using a dynamic programming method.

[0049] Specifically, based on the vehicle's real-time status, the onboard controller dynamically adjusts energy distribution using a rolling time-domain optimization strategy in each control cycle. Δt Inside, at the current moment t Starting with the boundary constraints of vehicle operating status data, in the local prediction time domain... Internally built with hydrogen fuel cell output power The minimum objective function is described as follows: (twenty one) In the formula: This is the weighting coefficient for the hydrogen consumption cost tracking error; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; The weighting coefficient for the vehicle operating efficiency tracking error; For time step; The weighting coefficient for the target speed of the vehicle; The primary prediction time domain; For the predicted health status of hydrogen fuel cells; For the first The predicted operating efficiency of the vehicle within each control cycle; The weighting coefficient for the target speed of the vehicle; For at any time Predicted vehicle speed; The target speed for the vehicle; The predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell; Specifically, after the vehicle driver starts the engine, the onboard controller detects and reads the current vehicle parameters and road information, and then... Figure 5 The vehicle dynamics model is obtained as follows: (twenty two) In the formula: air density; Road slope; For the efficiency of the transmission system; The derivative of the vehicle speed, i.e. ; For the first k The differential value of the vehicle speed within each control cycle, i.e., the instantaneous acceleration within that cycle. ; Indicates the first k The actual output power of the hydrogen fuel cell within each control cycle; Specifically, Indicates the first k The actual output power of the hydrogen fuel cell within each control cycle, i.e., the discrete quantity, and Representing continuous time t The corresponding output power. They are essentially the same, that is... ,in For the first k The starting time of each control cycle is presented in a discrete form in the vehicle dynamics model of this embodiment.

[0050] This embodiment employs a dynamic programming method in each control cycle. Δt Solving for the objective function that minimizes the output power of the hydrogen fuel cell, and combining it with the vehicle dynamics model, yields the power distribution command between the hydrogen fuel cell and the power battery: (twenty three) In the formula: For at any time t The optimal control command vector; For at any time t The optimal output power of a hydrogen fuel cell; For at any time t The optimal power battery output power; Preferably, the weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation is dynamically adjusted based on the current or temperature of the hydrogen fuel cell. When the current or temperature of the hydrogen fuel cell exceeds a set threshold, (twenty four) in, This indicates the safety threshold. The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; This represents the initial value for the hydrogen fuel cell lifespan weighting coefficient. The safety threshold for the health status of hydrogen fuel cells; This represents the lower threshold for the health status of hydrogen fuel cells. Specifically, when the on-board controller executes the energy distribution command for the hydrogen fuel cell, it introduces a hydrogen fuel cell lifespan coordination mechanism to adjust the current of the hydrogen fuel cell. or temperature When the threshold is exceeded, the weighting coefficient is dynamically adjusted.

[0051] In this embodiment, an energy management strategy for hydrogen fuel cell vehicles based on vehicle-road cooperative information is implemented. A vehicle-road cloud service simulation platform is constructed using the integrated wireless and transportation platform iTETRIS as its framework. Virtual simulation testing and real-vehicle testing are employed to verify the network service performance. The process includes the following steps: (1) Construct a vehicle-road cloud service simulation platform based on iTETRIS framework, integrating application (APP), integrated circuit design (iCS), traffic simulation module (Simulation of Urban Mobility, SUMO) and discrete event simulator (NetworkSimulator 3, NS3) to realize the simulation of vehicle, traffic and intermediate layer status and wireless communication data. Based on the simulation platform architecture, connect the vehicle controller through the on-board diagnostic port (On-board Diagnostic, OBD) to conduct virtual simulation test and detect the network performance impact after sending probe data packets.

[0052] Specifically, a vehicle-road-cloud service simulation platform based on the iTETRIS framework will be constructed, referencing... Figure 6 The simulation platform architecture is as follows: The vehicle-road cloud service simulation platform uses iTETRIS as its framework, connecting the cloud layer, middle layer, and bottom layer servers to the iCS module, which has APP management, simulation configuration, and hardware management functions, for data prediction processing and the execution of control command algorithm generation. It connects to the SUMO module and NS3 simulator via a simulator to simulate vehicle, traffic, and middle layer states, and expands the simulation functions for wireless communication data transmission and access. Furthermore, by inputting traffic information and environmental status information into the simulation platform, it enables comprehensive network performance testing and verification of onboard equipment and traffic facilities.

[0053] Based on a simulation platform architecture, virtual simulation testing was conducted by connecting the vehicle's OBD system to the on-board controller to detect the network performance impact after sending probe data packets. (Reference) Figure 7 The specific process for network performance testing is as follows: After receiving the network test command, the simulation platform sends test data packets to the vehicle controller of the vehicle under test. Upon receiving the returned data packets, it parses the messages sent and received by the vehicle controller, records communication metrics such as connection establishment time, latency, and packet loss rate, and records these data in the cloud's log file. Upon receiving the test completion command, the platform analyzes and saves the log file; otherwise, it continues to send test data packets, waiting for the test completion command.

[0054] (2) Build a real vehicle test platform to verify the vehicle's network communication and driving performance, referring to Figure 8It adopts a hierarchical and mesh-based configuration architecture. The hardware architecture is as follows: The scenario simulation layer verifies the network and driving performance of the vehicle under test by simulating the real operating conditions of the vehicle and the intermediate layer. The perception layer collects current vehicle, traffic, and environmental status information through monitoring equipment such as cameras and lidar at road crossings, while also providing data support for the service layer and scenario simulation layer. Simultaneously, the industrial control computer of the perception layer outputs early warning information to the roadside electronic display screens.

[0055] The network layer is used for information transmission between various layers. In order to meet the communication quality requirements of different protocol types and service needs, it adopts a variety of communication technologies such as Wi-Fi, 5G, LTE, and V2X. Each communication module is connected in the form of a fiber optic ring network. All devices on the test platform are connected to the network layer through communication interfaces and participate in the information interaction of the entire real vehicle test platform.

[0056] As the core component of the real vehicle testing platform, the service layer has the functions of data acquisition and processing and key distribution, and processes test environment data and device status transmitted from other layers of the network layer.

[0057] Simultaneously, based on this hardware architecture, a hierarchical and grid-based management configuration based on vehicle-road cooperative information is established to improve the system's environmental adaptability. The hierarchical management configuration ensures that the status information of all modules collected on the simulation platform can be recorded and stored, and that redundancy errors in one level do not affect other levels. During network testing, the service layer marks the corresponding location information in the log file formed by the communication indicator data records. The network performance testing computer is connected to the network layer and receives data information from all devices connected to the simulation platform through Wireshark software on the device for diagnosing network performance indicators.

[0058] A network-based management system is employed to configure different device terminals. Other network devices at different levels can be remotely configured via the network layer, and monitoring equipment and road test electronic displays can be configured via industrial control computers. For traffic lights in the scenario simulation layer, their proprietary protocols need to be converted to a unified protocol for the simulation platform. During testing, identifiers are marked on the vehicles, enabling the monitoring equipment in the perception layer to collect the status information of the tested vehicles and transmit it to the service layer via the wireless network of the network layer.

[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information, characterized in that, This includes a vehicle energy management system; the vehicle energy management system includes a top-level architecture, a middle-level architecture, and a bottom-level architecture; The underlying architecture is a hydrogen fuel cell vehicle; the hydrogen fuel cell vehicle includes an on-board controller and on-board actuators; the on-board controller is used to monitor and store vehicle operating status data in real time; The intermediate layer architecture consists of roadside units and traffic facilities; the roadside units and traffic facilities are used to acquire traffic status data and receive dispatch signals from the cloud-based traffic control platform. The top-level architecture is a cloud-based traffic control platform; the cloud-based traffic control platform includes a cloud data center and a cloud server; the cloud data center stores map data; the cloud server is used to obtain the predicted speed of the target hydrogen fuel cell vehicle using a long short-term memory network based on map data, vehicle operating status data, and traffic status data, and to obtain the predicted instantaneous hydrogen consumption rate, predicted fuel cell health status, and predicted operating efficiency of the hydrogen fuel cell using a multi-task LSTM prediction model, so as to obtain the optimal trajectory sequence of vehicle speed and the optimal output power command sequence of the fuel cell, and then obtain the power allocation command between the hydrogen fuel cell and the power battery, and send it to the hydrogen fuel cell vehicle for execution by the on-board execution equipment.

2. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information according to claim 1, characterized in that, The management method for a real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information includes the following steps: S1: Obtain vehicle operating status data of the target hydrogen fuel cell vehicle through the on-board controller of the target hydrogen fuel cell vehicle; S2: Using a cloud-based traffic control platform and a roadside unit and traffic facilities with an intermediate layer architecture, environmental event sequence data is acquired to obtain time series data with multidimensional characteristics of the target hydrogen fuel cell vehicle based on the vehicle operation status data of the target hydrogen fuel cell vehicle. S3: Based on the vehicle operating status data and environmental event sequence data of the target hydrogen fuel cell vehicle, a long short-term memory network is used to obtain the LSTM hidden state sequence, so as to obtain the vehicle driving condition time sequence in the future main prediction time domain, and then obtain the predicted speed of the target hydrogen fuel cell vehicle. S4: Based on the LSTM hidden state sequence, a multi-task LSTM prediction model is used to obtain the predicted instantaneous hydrogen consumption rate, the predicted health status of the fuel cell, and the predicted operating efficiency of the hydrogen fuel cell. S5: Based on the predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell, the predicted health status of the fuel cell, and the predicted operating efficiency, construct an objective function with the total cost as the optimization objective. Then, use model predictive control to obtain the optimal trajectory sequence of vehicle speed that minimizes the objective function and the optimal output power command sequence of the fuel cell. S6: Establish an objective function and vehicle dynamics model that minimizes the output power of the hydrogen fuel cell. Based on the optimal trajectory sequence of vehicle speed and the optimal output power command sequence of the hydrogen fuel cell, obtain the power allocation command between the hydrogen fuel cell and the power battery, i.e. the output power of the hydrogen fuel cell and the output power of the power battery, and send it to the underlying architecture for execution by the on-board execution device.

3. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information according to claim 2, characterized in that, The vehicle operating status data of the target hydrogen fuel cell vehicle includes: standardized vehicle speed, hydrogen storage capacity of the hydrogen storage tank, and state of charge of the power battery; and boundary constraints are set for the vehicle operating status data. The environmental event sequence data includes: the slope, curvature, traffic light status set, current charging price, and charging waiting time on the road along the target hydrogen fuel cell vehicle's driving trajectory.

4. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information according to claim 3, characterized in that, The S3 includes: a main state coding branch and a secondary state coding branch; S31: The main state encoding branch of the LSTM dual-branch structure prediction network is used to obtain the hidden state of the vehicle operating state sequence based on the vehicle operating state data of the target hydrogen fuel cell vehicle. S32: The sub-state encoding branch of the prediction network using the LSTM dual-branch structure is used to obtain the hidden state of the environmental event sequence based on the environmental event sequence data; S33: Obtain the merged LSTM hidden state sequence based on the hidden states of the vehicle operation state sequence and the hidden states of the environmental event sequence; S34: Based on the merged LSTM hidden state sequence, obtain the vehicle driving condition time series of the target hydrogen fuel cell vehicle in the future main prediction time domain, so as to obtain the predicted speed of the target hydrogen fuel cell vehicle.

5. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information according to claim 4, characterized in that, The predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell is expressed as follows: In the formula: The predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell; The mapping function is obtained through learning; This refers to the output power of the hydrogen fuel cell; Power required by the vehicle; SOC This refers to the state of charge of the power battery. The control cycle time for the operation of a hydrogen fuel cell; The predicted health status of the fuel cell is represented as follows: In the formula: For the predicted health status of fuel cells; The control cycle time for the operation of a hydrogen fuel cell; N The total number of control cycles for the operation of the hydrogen fuel cell; For the first The current of the power battery within each control cycle; This is the rated current of the hydrogen fuel cell; For the first Reference temperature of the hydrogen fuel cell within each control cycle; For the first The actual temperature of the power battery within each control cycle; The predicted operating efficiency is expressed as follows: In the formula: For the first k The predicted operating efficiency of the vehicle within each control cycle; , , All are weighting coefficients, satisfying ; , , These are the efficiency of the hydrogen fuel cell, the efficiency of the power battery, and the instantaneous efficiency of the motor.

6. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information according to claim 5, characterized in that, The objective function, which aims to minimize the total cost, is expressed as follows: In the formula: For at any time t The total cost optimization objective function; The primary prediction time domain; The control cycle time for the operation of a hydrogen fuel cell; This is the weighting coefficient for the hydrogen consumption cost tracking error; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; The weighting coefficient for the vehicle operating efficiency tracking error; For hydrogen fuel cells in the first k The output power within each control cycle is Instantaneous hydrogen consumption rate at time; For the first Equivalent factor for each control cycle; For the predicted health status of fuel cells; For the first The predicted operating efficiency of the vehicle within each control cycle; The weighting coefficient for the target speed of the vehicle; For at any time Predicted vehicle speed; The target speed for the vehicle; in, In the formula: For the first k Equivalent factor for each control cycle; The coefficient is positively correlated with the degree of fluctuation in the predicted vehicle speed; This serves as a reference value for the health status of hydrogen fuel cells. This is a transpose.

7. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information according to claim 6, characterized in that, The output power of the hydrogen fuel cell The minimum objective function is expressed as follows: In the formula: This is the weighting coefficient for the hydrogen consumption cost tracking error; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; The weighting coefficient for the vehicle operating efficiency tracking error; For time step; The weighting coefficient for the target speed of the vehicle; The primary prediction time domain; For the predicted health status of hydrogen fuel cells; For the first The predicted operating efficiency of the vehicle within each control cycle; The weighting coefficient for the target speed of the vehicle; For at any time Predicted vehicle speed; The target speed for the vehicle; This represents the predicted instantaneous hydrogen consumption rate of the hydrogen fuel cell.

8. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information according to claim 7, characterized in that, The vehicle dynamics model is represented as follows: In the formula: air density; Road slope; For the efficiency of the transmission system; The derivative of the vehicle speed, i.e. ; For the first k The differential value of the vehicle speed within each control cycle, i.e., the instantaneous acceleration within that cycle. ; Indicates the first k The actual output power of the hydrogen fuel cell within each control cycle.

9. The real-time energy management architecture for hydrogen fuel cell vehicles based on vehicle-road information as described in claim 8, characterized in that, The weighting coefficient for cost tracking error of hydrogen fuel cell lifespan degradation is dynamically adjusted based on the current or temperature of the hydrogen fuel cell. When the current or temperature of the hydrogen fuel cell exceeds a set threshold, in, Indicates the safety threshold; The weighting coefficient for the cost tracking error of hydrogen fuel cell lifespan degradation; This represents the initial value for the hydrogen fuel cell lifespan weighting coefficient. The safety threshold for the health status of hydrogen fuel cells; This represents the lower threshold for the health status of a hydrogen fuel cell.

Images