Driving range simulation method and apparatus for new energy vehicle, and medium and device

Abstract

A driving range simulation method and apparatus for a new energy vehicle, and a medium and a device. The method comprises: splitting a vehicle thermal management system model into four sub-models (110); establishing a training data set for each sub-model (120); establishing reduced-order sub-models for each sub-model (130); and integrating the model, and performing high- and low-temperature driving range simulation on a new energy vehicle (140).

Description

A method, apparatus, medium and equipment for simulating the driving range of new energy vehicles

[0001] This application claims priority to Chinese Patent Application No. 202411184611.6, filed on August 27, 2024, entitled "A Method, Apparatus, Medium and Equipment for Simulating the Driving Range of New Energy Vehicles", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application belongs to the field of automotive thermal management technology, and in particular relates to a method, device, medium and equipment for simulating the driving range of new energy vehicles. Background Technology

[0003] In summer cooling and winter heating conditions, the thermal management system of new energy vehicles consumes a large amount of electrical energy, resulting in a significant reduction in actual driving range and hindering the market competitiveness of new energy vehicles.

[0004] In related technologies, in order to improve the actual driving range of pure electric vehicles under high and low temperature conditions, the energy consumption of the thermal management system will be optimized based on the simulation results of high and low temperature driving range during the development stage of pure electric vehicles.

[0005] However, high and low temperature driving range simulation requires coupling the vehicle dynamics model and the thermal management system model. Since the thermal management system model converges slowly during the simulation process, the high and low temperature driving range simulation cycle is very long, resulting in low efficiency. Summary of the Invention

[0006] To address the technical problems mentioned above, this application provides a method, apparatus, medium, and equipment for simulating the driving range of new energy vehicles. The vehicle thermal management system model is decomposed into sub-models for the air conditioning system, motor system cooling circuit, battery system cooling circuit, and passenger compartment. Training datasets for each subsystem in the vehicle thermal management system are established based on one-dimensional thermal management simulation. Using machine learning methods, reduced-order sub-models for each subsystem in the vehicle thermal management system are trained using the training datasets. These reduced-order sub-models and management control strategies are integrated into the vehicle dynamics model, achieving efficient simulation of driving range at high and low temperatures.

[0007] To achieve the above objectives, this application adopts the following technical solution:

[0008] The first aspect of this application provides a method for simulating the driving range of new energy vehicles.

[0009] A method for simulating the driving range of new energy vehicles includes:

[0010] Obtain the parameters of the controlled component and the vehicle parameters corresponding to multiple consecutive time points as a sequence of input variables;

[0011] Based on the input variable sequence, the output variables corresponding to the multiple reduced-order sub-models are predicted through multiple reduced-order sub-models; based on the output variables corresponding to the multiple reduced-order sub-models, the load power and system state parameters are obtained; based on the load power, the vehicle parameters are calculated through the vehicle dynamics model; based on the system state parameters, the parameters of the controlled components are calculated through the management and control strategy.

[0012] When the battery is depleted, the driving range is calculated; when the battery is not depleted, the calculated parameters of the controlled components and the vehicle parameters are used as the latest data to update the input variable sequence, and the output variable is updated according to the updated input variable sequence.

[0013] The training process of the reduced-order sub-model includes: splitting the vehicle thermal management system model into an air conditioning system sub-model, a motor system cooling circuit sub-model, a battery system cooling circuit sub-model, and a passenger compartment model; generating a sequence of sample input variables for each sub-model; predicting sample output variables for the sequence of sample input variables through simulation; matching the sequence of sample input variables with the sample output variables to obtain a training dataset; and training the neural network using the training dataset to obtain the reduced-order sub-model.

[0014] Furthermore, the input variables of the air conditioning system sub-model include: compressor speed, condenser inlet air temperature, condenser inlet air volume, evaporator inlet air humidity, evaporator inlet air volume, evaporator inlet air temperature, electronic expansion valve superheat, thermal expansion valve superheat, inlet water temperature and inlet water flow of the open-loop battery circuit inlet plate heat exchanger; the output variables of the air conditioning system sub-model include: compressor power, compressor exhaust temperature, blower power, fan power, system refrigerant high pressure, system refrigerant low pressure, HVAC outlet air temperature, plate heat exchanger refrigerant outlet temperature and outlet pressure, and battery inlet water temperature.

[0015] Furthermore, the input variables of the motor system cooling circuit sub-model include: water pump speed, radiator air intake speed, radiator air intake temperature, motor speed, motor torque, and electronic control heat; the output variables of the motor system cooling circuit sub-model include: coolant flow rate, radiator inlet water temperature, motor controller inlet water temperature, motor body temperature, and electrical power of each component of the motor system.

[0016] Furthermore, the input variables of the battery system cooling circuit sub-model include: battery inlet water temperature, electric water pump speed, battery charge, current, voltage, and initial battery temperature; the output variables of the battery system cooling circuit sub-model include: coolant flow rate, battery inlet / outlet water temperature, maximum cell temperature, minimum cell temperature, and battery heat generation.

[0017] Furthermore, the input variables of the occupant cabin model include: air temperature and air volume at the air conditioning unit outlet, ambient temperature, solar radiation intensity, vehicle speed, and initial air temperature inside the cabin; the output variables of the occupant cabin model include: average air temperature inside the cabin and blower power.

[0018] Furthermore, training the neural network using a training dataset includes:

[0019] Initialize a population containing multiple individuals, each individual corresponding to a neural network and learnable parameters; the neural network corresponds to a number of hidden layers.

[0020] For each individual's corresponding neural network, the training dataset is used for training, the learnable parameters are optimized, and the error value is obtained after training is completed.

[0021] Fitness is determined based on the error value, and screening and genetic operations are performed on the population based on the fitness. The screening and genetic operations are used to determine the optimal individual.

[0022] If the maximum number of iterations is reached, the neural network trained based on the optimal individual is used as the reduced-order sub-model; if the maximum number of iterations is not reached, the number of iterations is incremented by one, and the process returns to train the neural network corresponding to each individual.

[0023] Furthermore, for the reduced-order sub-model, the number of time points in the input variable sequence used for output variable prediction is consistent with the number of hidden layers corresponding to the optimal individual.

[0024] The second aspect of this application provides a simulation device for the driving range of a new energy vehicle.

[0025] A new energy vehicle range simulation device includes:

[0026] The reduced-order sub-model generation module is configured to: decompose the vehicle thermal management system model into an air conditioning system sub-model, a motor system cooling circuit sub-model, a battery system cooling circuit sub-model, and a passenger compartment model; generate a sequence of sample input variables for each sub-model obtained from the decomposition; predict sample output variables for the sample input variable sequence through simulation; map the sample input variable sequence and the sample output variables one-to-one to obtain a training dataset; and train the neural network using the training dataset to obtain the reduced-order sub-model.

[0027] The data acquisition module is configured to acquire the parameters of the controlled component and the vehicle parameters corresponding to multiple consecutive time points as a sequence of input variables;

[0028] The prediction module is configured to: predict the output variables corresponding to the multiple reduced-order sub-models based on the input variable sequence; obtain the load power and system state parameters based on the output variables corresponding to the multiple reduced-order sub-models; calculate the vehicle parameters based on the load power using the vehicle dynamics model; and calculate the parameters of the controlled component based on the system state parameters using a management and control strategy.

[0029] The judgment module is configured to: calculate the driving range when the battery is depleted; and update the input variable sequence using the calculated parameters of the controlled component and the vehicle parameters as the latest data when the battery is not depleted, and update the output variable according to the updated input variable sequence.

[0030] A third aspect of this application provides a computer-readable storage medium.

[0031] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the new energy vehicle range simulation method described in the first aspect above.

[0032] The fourth aspect of this application provides a computer device.

[0033] A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps in the new energy vehicle range simulation method described in the first aspect above.

[0034] The beneficial effects of the technical solutions provided in this application include at least the following:

[0035] This application decomposes the vehicle thermal management system model into sub-models for the air conditioning system, motor system cooling circuit, battery system cooling circuit, and passenger compartment. Based on one-dimensional thermal management simulation, a training dataset for each subsystem in the vehicle thermal management system is established. Using machine learning methods, a reduced-order sub-model for each subsystem in the vehicle thermal management system is trained using the training dataset. The reduced-order sub-model and management control strategy are integrated into the vehicle dynamics model, achieving efficient simulation of high and low temperature range.

[0036] This application breaks down the vehicle thermal management system and develops reduced-order sub-models for subsystems such as the air conditioning system, motor system cooling circuit, battery system cooling circuit, and passenger compartment. This improves the engineering feasibility and model accuracy of the vehicle thermal management system and promotes the engineering application value of reduced-order models in the field of vehicle thermal management simulation.

[0037] This application decomposes the vehicle thermal management system model into sub-models for the air conditioning system, the motor system cooling circuit, the battery system cooling circuit, and the passenger compartment. For each sub-model, a sequence of sample input variables is generated for simulation to obtain sample output variables. A training dataset is then constructed and a neural network is trained to obtain a reduced-order sub-model. These reduced-order sub-models can more efficiently predict various parts of the thermal management system. During simulation, based on the input variable sequence, these reduced-order sub-models can quickly obtain the output variables, thereby obtaining the load power and system state parameters. This reduces the complex calculations and convergence time of traditional thermal management system models in simulation, and improves the efficiency of high and low temperature range simulation.

[0038] By acquiring the parameters of the controlled components and the vehicle at multiple consecutive time points, an input variable sequence is constructed. This sequence is then used in conjunction with a reduced-order sub-model, a vehicle dynamics model, and management control strategies to simulate the vehicle's operating state under different conditions. The driving range is calculated when the battery is depleted, and the input and output variable sequences are updated when the battery is not depleted. This dynamic update method can more accurately simulate the vehicle's state changes during actual driving. The solution provided in this application can more accurately assess the driving range of new energy vehicles under high and low temperature conditions, providing a more accurate basis for optimizing the energy consumption of the thermal management system. This helps to better optimize the thermal management system during the development phase and improve the actual driving range. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0040] Figure 1 is a flowchart of a new energy vehicle range simulation method shown in Embodiment 1 of this application;

[0041] Figure 2 is an example diagram of an air conditioning system sub-model shown in Embodiment 1 of this application;

[0042] Figure 3 is a flowchart illustrating the process of establishing a sub-model training dataset as shown in Embodiment 1 of this application;

[0043] Figure 4 is a data transfer example diagram of the thermal management model, vehicle dynamics model and thermal management control strategy shown in Embodiment 1 of this application;

[0044] Figure 5 is a schematic diagram of the structure of a computer device shown in Embodiment 4 of this application. Detailed Implementation

[0045] The present application will be further described below with reference to the accompanying drawings and embodiments.

[0046] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0047] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0048] It should be noted that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and apparatus according to various embodiments of this application. It should be noted that each block in a flowchart or block diagram may represent a module, segment, or portion of code, which may include one or more executable instructions for implementing the logical functions specified in the various embodiments. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than that shown in the drawings. For example, two consecutively represented blocks may actually be executed substantially in parallel, or they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the flowcharts and / or block diagrams, and combinations of blocks in the flowcharts and / or block diagrams, may be implemented using dedicated hardware-based apparatus to perform the specified functions or operations, or may be implemented using a combination of dedicated hardware and computer instructions.

[0049] Example 1

[0050] This embodiment provides a method for simulating the driving range of new energy vehicles.

[0051] This embodiment provides a method for simulating the driving range of new energy vehicles. It uses machine learning to develop a reduced-order model of the thermal management system (i.e., the reduced-order sub-model mentioned above). Based on the reduced-order model, an engineering-based method for simulating the driving range of new energy vehicles is established, which can significantly improve the efficiency of high and low temperature driving range simulation to meet the industry demand for rapid iteration of vehicle development.

[0052] This embodiment provides a method for simulating the driving range of new energy vehicles, as shown in Figure 1, including the following steps:

[0053] Step 110: Divide the vehicle thermal management system model into four sub-models.

[0054] That is, the vehicle thermal management system model is broken down into an air conditioning system sub-model, a motor system cooling circuit sub-model, a battery system cooling circuit sub-model, and a passenger compartment model.

[0055] The vehicle thermal management system model is a digital model used to simulate the thermal equilibrium state of new energy vehicles (or traditional vehicles). It abstracts the processes of heat generation, transfer, and exchange among various vehicle subsystems into a computable model. The subsystems included in the vehicle thermal management system model include the air conditioning system, motor system, battery system, and passenger compartment.

[0056] The vehicle thermal management system model is used to simulate the vehicle thermal management system. Simulation is a technical means of dynamically reproducing the operation process of the vehicle thermal management system in a computer based on the vehicle thermal management system model. It drives the model to calculate and simulate the changes in the thermal management state of the vehicle in real-world scenarios by inputting different operating parameters.

[0057] In a schematic representation, the vehicle thermal management system model is divided into four sub-models: air conditioning system, motor system cooling circuit, battery system cooling circuit, and passenger compartment, and data transmission interfaces are set between each sub-model.

[0058] For the data transfer interface between the sub-models, the relevant sub-model interaction parameters mainly include water flow rate, inlet and outlet water temperature, air volume, and air temperature, with specific parameters depending on the model. In the simulation of the vehicle thermal management system, the four sub-models (air conditioning system, motor cooling circuit, battery cooling circuit, and passenger compartment) do not operate independently; they must exchange key physical quantities, and the sub-model interaction parameters are the key physical quantities transferred between the sub-models.

[0059] The following sections will introduce the input variables and predicted variables (i.e., output variables) of the four sub-models: air conditioning system, motor system cooling circuit, battery system cooling circuit, and passenger compartment.

[0060] (1) Air conditioning system sub-model

[0061] Figure 2 shows a schematic diagram of an air conditioning system sub-model. As shown in Figure 2, the air conditioning system sub-model mainly includes an electric compressor 1, a blower 2, a condenser 5, an evaporator 3, a thermal expansion valve 4, an electronic expansion valve 6, a plate heat exchanger 7, an air conditioning circuit 8, and an open-loop battery water circuit 9.

[0062] For the air conditioning system sub-model, the main input variables include compressor speed, condenser inlet air temperature, condenser inlet air volume, evaporator inlet air humidity, evaporator inlet air volume, evaporator inlet air temperature, electronic expansion valve superheat, thermal expansion valve superheat, inlet water temperature and inlet water flow of the inlet plate heat exchanger in the open-loop battery circuit, etc.

[0063] For illustrative purposes, compressor speed refers to the rotational speed of the compressor spindle; condenser inlet air temperature refers to the temperature of outside air entering the condenser; condenser inlet air volume refers to the air volume flowing through the condenser; evaporator inlet air humidity refers to the humidity of the air before entering the evaporator; evaporator inlet air volume refers to the air volume flowing through the evaporator; electronic expansion valve superheat refers to the superheat setting value of the electronic expansion valve; thermal expansion valve superheat refers to the superheat setting value of the thermal expansion valve; open-loop battery circuit inlet plate heat exchanger inlet water temperature refers to the temperature of the coolant entering the plate heat exchanger; open-loop battery circuit inlet plate heat exchanger inlet water flow rate refers to the flow rate of coolant flowing through the plate heat exchanger.

[0064] For the air conditioning system sub-model, the predicted variables (i.e., output variables) mainly include compressor power, compressor discharge temperature, blower power, fan power, system refrigerant high pressure, system refrigerant low pressure, HVAC (Heating, Ventilation, and Air Conditioning) outlet air temperature, plate heat exchanger refrigerant outlet temperature and outlet pressure, battery inlet water temperature, etc.

[0065] For illustrative purposes, compressor power refers to the compressor's electrical power; compressor discharge temperature refers to the coolant temperature at the compressor outlet; blower power refers to the blower's electrical power; fan power refers to the condenser fan's electrical power; system refrigerant high pressure refers to the coolant pressure on the discharge side; system refrigerant low pressure refers to the coolant pressure on the suction side; HVAC outlet air temperature refers to the air temperature supplied to the passenger compartment; plate heat exchanger refrigerant outlet temperature refers to the temperature of the coolant leaving the plate heat exchanger; plate heat exchanger refrigerant outlet pressure refers to the pressure of the coolant leaving the plate heat exchanger; and battery inlet water temperature refers to the temperature of the coolant entering the battery.

[0066] The design of the input and output variables of the above-mentioned air conditioning system sub-model enables refined modeling of the air conditioning system sub-model. This refined modeling allows the reduced-order sub-model to more accurately predict the energy consumption items of the air conditioning system itself (compressor, fan, blower power), and ensures accurate calculation of the cooling / heating effect (battery inlet water temperature) transferred to the battery circuit through the plate heat exchanger. This significantly improves the prediction accuracy of the air conditioning system sub-model under complex coupled conditions, thereby more realistically reflecting the impact of the air conditioning system on the vehicle's energy consumption and range, and supporting more effective thermal management control strategy optimization.

[0067] (2) Sub-model of motor system cooling circuit

[0068] The motor system cooling loop sub-model mainly includes an electric water pump, a motor, a motor controller, and a radiator. The input variables for the motor system cooling loop sub-model mainly include water pump speed, radiator inlet air velocity, radiator inlet air temperature, motor speed, motor torque, and electrical control heat. The predicted variables (i.e., output variables) mainly include coolant flow rate, radiator inlet water temperature, motor controller inlet water temperature, motor body temperature, and the electrical power of each component in the motor system.

[0069] Indicatively, the pump speed indicates how many revolutions the electronic pump makes per unit time; the radiator inlet air velocity refers to the average wind speed entering the radiator; the radiator inlet air temperature refers to the air temperature entering the radiator fins; the motor speed refers to the rotational angular velocity of the motor shaft; the motor torque refers to the torque output by the motor shaft, which determines the current; and the electrical control heat refers to the heat loss of the motor controller (such as an inverter).

[0070] Coolant flow rate refers to the flow rate of coolant in the circuit; radiator inlet temperature refers to the temperature of coolant when it enters the radiator; motor controller inlet temperature refers to the temperature of coolant when it enters the controller; motor body temperature usually refers to the average temperature of the motor housing; the electrical power of each component of the motor system refers to the electrical power of the electric water pump, motor, motor controller, radiator, etc., used for vehicle energy consumption statistics.

[0071] The design of the input variables of the above-mentioned motor system cooling circuit sub-model fully describes the working state of the motor cooling circuit. The design of the output variables enables the reduced-order model to directly output the main energy consumption sources of the motor cooling circuit (such as the electric power of the water pump and the cooling fan), ensuring that the energy consumption of the motor cooling circuit can be efficiently and accurately integrated into the whole vehicle simulation.

[0072] (3) Battery system cooling circuit sub-model

[0073] The battery system cooling circuit sub-model mainly includes an electric water pump and the battery. The input variables of the battery system cooling circuit sub-model mainly include the battery inlet water temperature, electric water pump speed, battery charge, current, voltage, and the initial temperature of the battery; the predicted variables (i.e., output variables) mainly include coolant flow rate, battery inlet / outlet water temperature, maximum cell temperature, minimum cell temperature, and heat generation of the battery.

[0074] For illustrative purposes, the battery inlet water temperature refers to the temperature of the coolant before it enters the battery pack; the electronic water pump speed refers to the speed of the water pump in the battery circuit; the battery charge refers to the current remaining usable capacity; the current refers to the real-time charging and discharging current; the voltage refers to the battery terminal voltage; and the initial battery temperature refers to the average temperature of the battery cells at the start of the simulation.

[0075] Coolant flow rate refers to the flow rate of coolant in the battery circuit; battery inlet / outlet water temperature refers to the temperature of coolant entering and leaving the battery pack, respectively; maximum cell temperature refers to the maximum temperature of the battery pack; minimum cell temperature refers to the minimum temperature of the battery pack; battery heat generation refers to the total heat power generated by the battery during charging and discharging.

[0076] The input variables of the aforementioned battery system cooling circuit sub-model cover key factors of battery heat generation and dissipation. The output variables include the cell's maximum / minimum temperature and the battery's heat generation. Accurately outputting the battery's heat generation is the direct basis for calculating battery temperature rise and cooling system load power, directly impacting energy consumption calculations. This precise modeling of the battery system cooling circuit sub-model improves the simulation accuracy and efficiency of battery thermal management, the aspect with the greatest impact on driving range. Accurate temperature and heat generation predictions are crucial for optimizing battery cooling / heating strategies, reducing battery thermal management energy consumption, and ultimately improving driving range at both high and low temperatures.

[0077] (4) Crew cabin model

[0078] The passenger compartment model mainly includes the air ducts and the passenger compartment. The input variables of the passenger compartment model mainly include the air temperature and air volume at the air conditioning unit outlet, the ambient temperature, the solar radiation intensity, the vehicle speed, and the initial air temperature inside the compartment; the predicted variables (i.e., the output variables) mainly include the average air temperature inside the compartment and the blower power.

[0079] For illustrative purposes, the air temperature at the air conditioning unit outlet refers to the temperature of the air supplied into the cabin; the air volume at the air conditioning unit outlet refers to the volume of air supplied into the cabin; the ambient temperature refers to the temperature of the atmosphere surrounding the vehicle; the vehicle speed refers to the speed of the vehicle; and the initial temperature of the cabin air refers to the average temperature of the air in the passenger compartment at the start of the simulation.

[0080] The average cabin air temperature refers to the average temperature of the passenger compartment, which directly reflects the passengers' temperature perception; the blower power refers to the electrical power consumed by the blower motor.

[0081] The input variables of the aforementioned passenger cabin model fully define the cabin's thermal environment, while the output variables include the average cabin air temperature (measuring comfort) and blower power (direct energy consumption). This accurate modeling of the passenger cabin makes the simulation of passenger cabin heat load and comfort maintenance energy consumption more realistic and reliable. It helps optimize air conditioning control strategies (such as airflow and temperature settings), minimizing the energy consumption of components like the blower while ensuring basic comfort, thereby contributing to increased driving range.

[0082] Step 120: Establish the training dataset for each sub-model.

[0083] In some embodiments, the range of input variable parameters is determined, and training datasets corresponding to the four subsystem models of the vehicle thermal management system are established based on one-dimensional simulation software.

[0084] When determining the range of input variables, it is essential to cover the actual simulation conditions to avoid extrapolation reducing model accuracy. Simultaneously, the range of input variables should not be too large to prevent an increase in the dataset size and training time. Illustratively, extrapolation refers to the phenomenon where a reduced-order submodel predicts input parameters exceeding the training data range, resulting in a significant decrease in prediction accuracy because it has not learned the patterns of that range. Therefore, the range of input variables must fully encompass the boundaries of the operating conditions that vehicles may encounter in actual operation to ensure the reliability of the model's predictions in real-world application scenarios.

[0085] Based on the input variables of each subsystem identified above, and by determining the range of values ​​for the input variables based on experience.

[0086] Optionally, training datasets are established based on one-dimensional simulation software for the four subsystem models corresponding to the vehicle thermal management system. This includes: generating a sequence of sample input variables for each sub-model obtained from the splitting; predicting the sample output variables for the sample input variable sequence through simulation; and matching the sample input variable sequence with the sample output variables to obtain the training dataset.

[0087] Optionally, for each sub-model obtained from the decomposition, a sequence of sample input variables is generated. That is, the sequence of sample input variables for each sub-model is determined based on the range of input variable parameters corresponding to each sub-model. For example, for a specific sub-model (such as an air conditioning system sub-model), multiple sets of consecutive time-series input parameter combinations are designed based on the physical range of its input variables (such as compressor speed range of 1000-6000 rpm) to form a sequence of sample input variables. For instance, 1000 sets of sample input variable sequences are generated for the air conditioning system sub-model, each set containing parameters for five consecutive time periods, such as "compressor speed, condenser inlet air temperature, and evaporator inlet air humidity."

[0088] Optionally, each sub-model corresponds to one or more sequences of sample input variables. The sequence of sample input variables includes sample input variables corresponding to multiple time points.

[0089] The sample input variables include controlled component parameters and vehicle parameters. Controlled component parameters refer to the directly adjustable operating parameters of components in the vehicle's thermal management system, while vehicle parameters are those reflecting the overall vehicle operating status. Illustratively, the controlled component parameters and vehicle parameters are also the parameters in the input variables of the aforementioned sub-models. For example, in the air conditioning system sub-model, the controlled component parameters are compressor speed and electronic expansion valve superheat, while the vehicle parameters are condenser inlet air temperature and condenser inlet air volume.

[0090] Optionally, through simulation, the sample output variables are predicted for the sample input variable sequence. A training dataset is obtained by mapping the sample input variable sequence and the sample output variables one-to-one. This means that the sample input variable sequence is input into a sub-model, the simulation is run, and the parameters output by the sub-model are used as the sample output variables. Then, the simulated sample output variables are associated one-to-one with the sample input variable sequence to form a training dataset, which is used to subsequently train the reduced-order sub-model (such as a neural network), allowing the reduced-order sub-model to learn the mapping relationship between input and output.

[0091] Please refer to Figure 3. In some embodiments, the steps for building a training dataset include the following:

[0092] Step 121: Randomly generate a set of input variables within the range of values.

[0093] First, generate several random numbers within the range of input variable values. The number of random numbers determines the amount of data in the training set, which is generally more than 10,000.

[0094] To illustrate, for each sub-model's input variables (such as compressor speed and condenser inlet air temperature in an air conditioning system), within their defined physical range, several sets of random values ​​are generated using a computer program (such as MATLAB's random functions). For example, 10,000 sets of input variables are generated for the air conditioning system sub-model, each set containing specific values ​​for all input parameters such as "compressor speed, condenser inlet air temperature, and evaporator inlet air humidity".

[0095] Optionally, random sampling should follow a uniform or normal distribution (depending on the actual operating conditions) to ensure coverage of both extreme operating conditions (such as high temperature + high speed) and normal operating conditions (such as normal temperature + medium speed).

[0096] Step 122: Conduct sub-model simulation based on the input variable set.

[0097] Secondly, based on the coupling of MATLAB Simulink with one-dimensional thermal management simulation software (such as KULI, AMESim and GT), simulations of each sub-model are carried out. During the simulation, the input variables are controlled to change over time to obtain the simulation results. That is, the input of each sub-model is a sequence that changes over time.

[0098] For illustrative purposes, the input variables are not static values, but rather sequences that change over time (i.e., a sequence of sample input variables). For example, the motor speed in the cooling circuit of the motor system needs to simulate a dynamic process from 0→10000rpm→5000rpm (corresponding to the vehicle's start-up→acceleration→cruising conditions), rather than a fixed speed. The simulation software calculates the changes in output variables in real time based on the dynamic input. For instance, in the air conditioning system, as the compressor speed increases from 2000rpm to 5000rpm, the compressor power, outlet air temperature, and other outputs will dynamically adjust over time. The simulation results will record the complete time-series data of this process as the simulation outcome.

[0099] It should be noted that if there is a coupling relationship between sub-models (such as an air conditioning system cooling the battery through a plate heat exchanger), interactive parameters (such as battery inlet water temperature and water flow rate) need to be transmitted through a data interface during simulation to ensure the realism of the dynamic simulation.

[0100] Step 123: Analyze the simulation results to obtain the sub-model order reduction training dataset.

[0101] Finally, based on the simulation results of the one-dimensional simulation software, the outputs under different sample input variable sequences are obtained, and after sorting, the training datasets of the corresponding sub-models are obtained.

[0102] Specifically, automated scripts are developed to extract the sequence of sample input variables and the values ​​of output variables of interest (i.e., sample output variables) from the simulation results file. The sequence of sample input variables consists of input parameters designed for several consecutive time points, while the sample output variables are determined according to the development requirements of the reduced-order model. Generally, the output variables of the air conditioning system include parameters such as compressor power, compressor exhaust temperature, blower power, fan power, system refrigerant high pressure, system refrigerant low pressure, HVAC outlet air temperature, plate heat exchanger refrigerant outlet temperature and outlet pressure, and battery inlet water temperature; the output variables of the motor system cooling circuit include coolant flow rate, radiator inlet water temperature, motor controller inlet water temperature, motor body temperature, and electrical power of various components of the motor system; the output variables of the battery system cooling circuit mainly include coolant flow rate, battery inlet / outlet water temperature, maximum cell temperature, minimum cell temperature, and battery heat generation.

[0103] The input variable sequence and output variable values ​​generated by the automated script are arranged in order in an Excel file as the training dataset.

[0104] Step 130: Establish the reduced-order sub-models of each sub-model.

[0105] That is, the neural network is trained using the training dataset to obtain a reduced-order sub-model.

[0106] In a schematic manner, an appropriate algorithm is selected, and the neural network is trained using a training dataset to obtain a reduced-order model (i.e., a reduced-order sub-model) of the four subsystems of the thermal management system.

[0107] Optionally, different modeling models can be selected to establish reduced-order sub-models based on the heat capacity of the subsystem. Illustratively, heat capacity refers to the system's ability to store heat. Systems with large heat capacity absorb or release heat, resulting in slow temperature changes and requiring a longer time to transition from one state to another steady state (i.e., a long "transient process"); systems with small heat capacity respond quickly to temperature changes, have short transient processes, and can quickly reach a steady state.

[0108] Optionally, the modeling can include steady-state and transient models. Illustratively, steady-state models can ignore the time dimension, focusing only on the input-output relationship after the system reaches a steady state. Transient models require the time dimension, tracking the real-time response of the system from its initial state to a steady state (or during dynamic changes), focusing on the continuous changes of parameters over time.

[0109] Among them, the cooling circuits of the air conditioning system and motor system have relatively small system heat capacity, so the influence of the time domain on the model is small and the influence of the transient stage can be ignored. A reduced-order model can be established using a steady-state model.

[0110] Among them, the battery system cooling circuit and the passenger compartment model have large heat capacities, and the transient characteristics of the model cannot be ignored. Therefore, a transient model is needed to establish a reduced-order model.

[0111] In this embodiment, the development of reduced-order models for each subsystem can be accomplished using the software Altair ROMAI, which can train both steady-state and transient models. Alternatively, the NSS (neural state-space model) model can also be used to train the reduced-order models.

[0112] Optionally, the reduced-order model for each subsystem can employ an LSTM (Long Short-Term Memory) network. As mentioned above, different sub-models are affected by the time domain to varying degrees. This embodiment proposes using a genetic algorithm to optimize the number of hidden layers in different sub-models to obtain the reduced-order sub-models. Specifically:

[0113] (1) Initialize the population. The population contains multiple individuals, each of which corresponds to a neural network and learnable parameters.

[0114] The neural network has a number of hidden layers.

[0115] To illustrate, for a certain sub-model, the population and number of iterations are initialized. The population contains several individuals, each of which encodes the number of hidden layers and initial learnable parameters of an LSTM network.

[0116] Illustratively, the population is the set of objects to be optimized by the genetic algorithm, consisting of multiple individuals. Each individual corresponds to a complete LSTM network configuration, containing two key elements:

[0117] LSTM network structure (i.e., the neural network mentioned above): The core is the number of hidden layers (such as 1 layer, 2 layers, 3 layers, etc., which needs to be preset within a reasonable range according to the characteristics of the sub-model. For example, the battery system cooling circuit sub-model with strong time domain influence can be set with 2-5 layers, and the occupant cabin sub-model with weak time domain influence can be set with 1-3 layers); In addition, it can also include the number of neurons in the hidden layers, the type of activation function, etc. (but this embodiment focuses on optimizing the number of hidden layers).

[0118] Learnable parameters include the weights of the LSTM network (such as the connection weights between the input layer and the hidden layer, between hidden layers, and between the hidden layer and the output layer) and biases, which are generated through random initialization (such as random values ​​that follow a normal distribution).

[0119] Optionally, since different sub-models are affected by the time domain to varying degrees (e.g., battery system temperature changes have a strong lag and a large time domain impact; passenger cabin temperature response is faster and a weak time domain impact), the search range for the number of hidden layers needs to be set separately for each sub-model when initializing the population. For example:

[0120] Battery system cooling circuit sub-model: The number of hidden layers is set to 3-5 layers (to adapt to strong time-domain dependencies);

[0121] Crew cabin model: The number of hidden layers is set to 1-2 layers (to adapt to weak temporal dependencies).

[0122] (2) For each individual's corresponding neural network, the training dataset is used for training, the learnable parameters are optimized, and the error value is obtained after training is completed.

[0123] To illustrate, for each individual, the LSTM network is trained using the training dataset to optimize the learnable parameters. After training, the error value of the LSTM network for each individual is obtained.

[0124] Indicatively, for the currently optimized sub-model (such as an air conditioning system sub-model), sample input variable sequences and sample output variables are used as training data. The LSTM network training process includes: 1) inputting the sample input variable sequence into the LSTM network corresponding to each individual; 2) calculating the network's predicted output (i.e., the predicted value of the sub-model's output variable) through forward propagation; 3) comparing the difference between the predicted output and the sample output variable, and iteratively optimizing the network's learnable parameters (weights and biases) through backpropagation to minimize the prediction error. After training, the error value of the LSTM network on the validation set (split from the training dataset to evaluate generalization ability) is calculated using preset error metrics (such as mean squared error and mean absolute error). The smaller the error value, the more accurately the LSTM network fits the input-output relationship of the current sub-model. Each individual corresponds to an error value; the smaller the error value, the better the individual's performance.

[0125] (3) Determine the fitness based on the error value, and perform population screening and genetic operations based on the fitness.

[0126] Screening and genetic manipulation are used to determine the optimal individual.

[0127] Among them, the error value and fitness are negatively correlated, that is, the smaller the error value, the higher the fitness.

[0128] Optionally, for the selection operation, individuals with high fitness are selected from the current population and retained for the next generation. For example, the top 30% of individuals with the highest fitness are directly retained.

[0129] Optionally, for genetic operations, these operations include crossover and mutation. Crossover: Two pairs of "parent individuals" are randomly selected from the selected superior individuals, and "offspring individuals" are generated through a crossover operation. For example, crossover can be performed on the number of hidden layers: Parent 1 has 3 hidden layers, Parent 2 has 2 hidden layers, and the offspring can inherit 3 layers (retaining the advantages of Parent 1) or 2 layers (combining the simplicity of Parent 2), or the learnable parameters of the parents can be fused through parameter crossover. Mutation: Some features of the offspring individuals are randomly fine-tuned (such as changing the number of hidden layers with low probability, or adjusting some learnable parameters). For example, if all individuals have 2 hidden layers, mutation may produce individuals with 3 hidden layers, exploring a better structure.

[0130] Optionally, individuals obtained from screening and genetic operations in the current iteration will be used in the next iteration, and together they will constitute the population in the next iteration.

[0131] Optionally, a population selection operation is performed based on fitness; genetic operations are then performed on the selected individuals to obtain the population for the next iteration. Alternatively, a population selection operation is performed based on fitness, and genetic operations are then performed on the population; the selected individuals and the genetically generated individuals are used as the population for the next iteration.

[0132] In some embodiments, the error value is used as a fitness function for population screening and genetic operations.

[0133] (4) When the number of iterations reaches the maximum number of iterations, the neural network trained based on the optimal individual is used as the reduced-order sub-model; when the number of iterations does not reach the maximum number of iterations, the number of iterations is incremented by one, and the neural network training corresponding to each individual is returned.

[0134] Indicatively, determine whether the number of iterations has reached the maximum number of iterations. If so, use the LSTM network trained based on the optimal individual as the reduced-order sub-model; otherwise, increment the number of iterations by 1 and return to step (2).

[0135] Indicatively, a maximum number of iterations is preset, with each iteration corresponding to a complete training and evolution in steps (2)-(3). When the number of iterations reaches the maximum, the individual with the highest fitness (smallest error) is selected from the current population as the optimal individual, and its corresponding LSTM network is the optimized optimal network. The number of hidden layers in this network is the optimal solution that adapts to the temporal characteristics of the current sub-model (e.g., the battery sub-model ultimately determines 3 hidden layers, and the crew cabin sub-model determines 2 layers). Then, the LSTM network corresponding to the optimal individual is used as the reduced-order sub-model of the current sub-model. If the number of iterations does not reach the maximum value, the number of iterations is incremented by 1, and step (2) is returned to. The training, evaluation, and evolution process is repeated using a new generation of population until the termination condition is met.

[0136] Optionally, the number of time points in the input variable sequence used for output variable prediction is consistent with the number of hidden layers corresponding to the optimal individual.

[0137] Number of time points in the input variable sequence: The number of consecutive time points contained in the input variable sequence used to predict the output variable. For example, if the input is "compressor speed, condenser temperature, etc. at times t1, t2, and t3", then the number of time points is 3.

[0138] The number of hidden layers corresponding to the optimal individual: where the optimal individual refers to the neural network with the best performance selected by the genetic algorithm (whose learnable parameters and structure have been optimized); the number of hidden layers refers to the number of hidden layers in the neural network (such as 1, 2 or 3 hidden layers).

[0139] In other words, when predicting the output variable later, the number of time points in the input variable sequence needs to be consistent with the number of hidden layers corresponding to the optimal individual. For example, if the neural network of the optimal individual has 3 hidden layers, it means that the network can extract the features of the input data most accurately through 3 layers of processing; at this time, the input sequence uses 3 time points, which means that the data contains the temporal relationship of 3 time points (such as the dynamic influence of t1→t2→t3), which corresponds exactly to the feature extraction capability of 3 hidden layers. Each hidden layer can specifically learn the feature interaction of different time points (such as the first layer processing the relationship between t1 and t2, the second layer processing the relationship between t2 and t3, and the third layer combining the features of the three layers to output the prediction result).

[0140] In the above scheme, a population containing various neural network structures (different numbers of hidden layers) and learnable parameters is initialized. Each individual is trained and the error is calculated. Then, the individual with the best fitness is selected based on the error. Through the global search capability of the genetic algorithm, the optimal solution for fitting the input-output relationship of a specific sub-model (such as an air conditioner or motor cooling system) is found from a large number of potential neural network structures and parameter combinations. This more accurately captures the nonlinear characteristics of the sub-model, significantly reduces the prediction error, and the output variable of the final reduced-order sub-model is closer to the true value.

[0141] Furthermore, genetic algorithms, through iterative selection and genetic operations (such as crossover and mutation), not only optimize the learnable parameters of the neural network but also adaptively adjust the network structure (such as the number of hidden layers). This allows the final reduced-order sub-model to have a more streamlined structure (avoiding redundant hidden layers) and more efficient parameters (reducing ineffective weights) while maintaining accuracy, thereby significantly reducing the computational cost of a single prediction.

[0142] Step 140: Integrate the model and conduct high and low temperature driving range simulation of new energy vehicles.

[0143] Optionally, a reduced-order sub-model of the four subsystems of the integrated vehicle thermal management system and the vehicle dynamics model are integrated to establish an integrated visualization simulation platform and conduct high and low temperature driving range simulation of new energy vehicles.

[0144] Indicatively, a high and low temperature driving range simulation model for new energy vehicles is established by integrating four reduced-order sub-models of the vehicle thermal management system, the vehicle dynamics model, and the thermal management system control strategy (strategy module) on an integrated platform (such as MATLAB Simulink).

[0145] Specifically, the trained reduced-order sub-models are output in the form of FMUs, and the reduced-order sub-models are called through the Simulink FMU module. Based on Simulink, the reduced-order sub-models are integrated with the thermal management system control strategy and the vehicle dynamics model, and simulation of the high and low temperature driving range of new energy vehicles is carried out.

[0146] During range simulation, the data transfer between the vehicle thermal management system, the vehicle dynamics model, and the control strategy is shown in Figure 4. Once the battery is depleted, the vehicle's remaining range at that moment is obtained, serving as the overall vehicle range under high / low temperature conditions. Illustratively, in Figure 4, after setting simulation boundaries in the thermal management model 401 (i.e., the vehicle thermal management system model) and the vehicle dynamics model 402, the range simulation is initiated. During the simulation, the thermal management model 401 transmits load power (such as air conditioning compressor power, electric water pump power, and radiator fan power) to the vehicle dynamics model 402, while the vehicle dynamics model 402 transmits vehicle speed and heat generation to the thermal management model 401. The thermal management model 401 transmits system state parameters (such as current battery temperature, maximum / minimum temperature difference of battery cells, actual temperature of the passenger compartment, coolant inlet and outlet temperatures, and refrigerant pressure) to the thermal management control strategy (module) 403, while the thermal management control strategy (module) 403 transmits controlled component parameters (such as compressor target speed, electric water pump target flow rate, fan on / off status, and expansion valve opening) to the thermal management model 401. The thermal management model 401 adjusts the operating parameters of the corresponding components of the vehicle based on the controlled component parameters.

[0147] Among them, the simulation boundary refers to the basic environment and initial state of the simulation. Schematic, the simulation boundary includes: (1) Environmental parameters: specific temperatures under high and low temperature conditions (e.g., -25℃ low temperature, 40℃ high temperature), humidity, solar radiation intensity, etc. (2) Initial state: initial battery charge (e.g., fully charged state), initial vehicle temperature (e.g., cold start state at low temperature), etc. (3) Driving conditions: preset driving cycle, specified vehicle speed change curve over time, etc.

[0148] Once the simulation boundaries are set, the simulation system can simulate the entire process of a vehicle going from full charge to depletion of power based on these boundaries, and ultimately calculate the driving range.

[0149] During the model integration process, it is necessary to properly set up the data transmission interfaces between various subsystems and between the subsystems and the vehicle dynamics model.

[0150] Optionally, the thermal management model 401 includes reduced-order sub-models corresponding to the air conditioning system, motor system, battery system, and passenger compartment, that is, pre-trained reduced-order sub-models corresponding to the air conditioning system sub-model, motor system cooling circuit sub-model, battery system cooling circuit sub-model, and passenger compartment model, respectively.

[0151] Specifically, the steps of the above-mentioned battery life simulation include:

[0152] (1) Obtain the parameters of the controlled component and the vehicle parameters (vehicle speed, heat generation, etc.) corresponding to multiple consecutive time points as the input variable sequence; based on the input variable sequence, predict the output variables corresponding to multiple reduced-order sub-models through multiple reduced-order sub-models.

[0153] To illustrate, the input variable sequence is fed into four reduced-order sub-models, and the sub-models output the corresponding output variables based on the trained neural network (such as an LSTM network).

[0154] The input variables for different reduced-order sub-models are different, and the output variables for different reduced-order sub-models are different.

[0155] For example, the input variables of the reduced-order sub-model corresponding to the input air conditioning system sub-model are compressor speed, condenser inlet air temperature, condenser inlet air volume, evaporator inlet air humidity, evaporator inlet air volume, evaporator inlet air temperature, electronic expansion valve superheat, thermal expansion valve superheat, and inlet water temperature and flow rate of the plate heat exchanger at the inlet of the open-loop battery circuit; the output variables include: compressor power, compressor exhaust temperature, blower power, fan power, system refrigerant high pressure, system refrigerant low pressure, HVAC outlet air temperature, plate heat exchanger refrigerant outlet temperature and outlet pressure, and battery inlet water temperature.

[0156] The input variables of the reduced-order sub-model corresponding to the input motor system cooling circuit sub-model are water pump speed, radiator inlet air velocity, radiator inlet air temperature, motor speed, motor torque, and electronic control heat; the output variables include: coolant flow rate, radiator inlet water temperature, motor controller inlet water temperature, motor body temperature, and electrical power of each component of the motor system.

[0157] The input variables of the reduced-order sub-model corresponding to the input battery system cooling circuit sub-model are battery inlet water temperature, electric water pump speed, battery charge, current, voltage, and battery initial temperature; the output variables include: coolant flow rate, battery inlet / outlet water temperature, maximum cell temperature, minimum cell temperature, and battery heat generation.

[0158] The input variables for the reduced-order sub-model corresponding to the passenger compartment sub-model are the air temperature and air volume at the air conditioning unit outlet, the ambient temperature, the solar radiation intensity, the vehicle speed, and the initial temperature of the air inside the compartment; the output variables include the average temperature of the air inside the compartment and the blower power.

[0159] (2) Based on the output variables corresponding to multiple reduced-order sub-models, the load power and system state parameters are obtained; based on the load power, the vehicle parameters are calculated through the vehicle dynamics model; based on the system state parameters, the parameters of the controlled components are calculated through the management and control strategy.

[0160] For load power: Summarize the energy consumption parameters output by each sub-model (such as compressor power, water pump power, fan power, etc.) to obtain the total energy consumption power (auxiliary energy consumption) of the thermal management system. Input the load power into the vehicle dynamics model, and calculate the drive energy consumption by combining the vehicle driving resistance (related to vehicle speed), thereby obtaining the real-time discharge current and remaining charge of the battery, and updating the vehicle parameters for the next moment (such as the vehicle speed and motor heat generation based on the driving conditions).

[0161] For system state parameters: extract key states from the sub-model output (such as battery temperature, cabin temperature, motor controller temperature, etc.) to reflect the real-time operating status of the thermal management system. Input the system state parameters into the thermal management control strategy module. The strategy module calculates the optimal adjustment parameters for the next moment (such as increasing compressor speed, increasing water pump flow) based on the target (such as maintaining battery temperature at 25°C and cabin temperature at 22°C), which are the new parameters of the controlled components.

[0162] (3) When the battery is depleted, the driving range is calculated; when the battery is not depleted, the calculated parameters of the controlled components and the parameters of the whole vehicle are used as the latest data to update the input variable sequence, and the output variable is updated according to the updated input variable sequence.

[0163] Indicatively, it determines whether the battery is depleted. If so, it obtains the vehicle's range by reading the simulation model's result file, which serves as the overall vehicle range under high / low temperature conditions.

[0164] Otherwise, the controlled component parameters and vehicle parameters obtained in step (2) are used as the latest data and added to the input variable sequence. Through each reduced-order sub-model, the output variables are obtained and returned to step (2) until the complete driving range is calculated.

[0165] In this approach, if the reduced-order sub-model uses a neural network LSTM, the input variable sequence is pruned based on the optimized number of hidden layers. Specifically, if the number of hidden layers is n, only the input variables from the latest n time steps are retained. This pruning ensures the timeliness and relevance of the input sequence, improving prediction accuracy.

[0166] This embodiment provides a method for simulating the driving range of new energy vehicles. The thermal management system is divided into several sub-models, such as the air conditioning circuit, the electric drive water circuit, the battery water circuit, and the passenger compartment. Training datasets for each thermal management system subsystem are established based on one-dimensional thermal management simulation. Machine learning methods are used to train the datasets to obtain reduced-order models of each thermal management system subsystem. The reduced-order models are integrated into the dynamic model to achieve efficient simulation of driving range at high and low temperatures.

[0167] This embodiment provides a method for simulating the driving range of new energy vehicles, which can efficiently conduct driving range simulation. Compared with conventional methods, it can shorten the simulation time from 5 hours to 5 minutes, significantly reducing the simulation time for high and low temperature driving range, and realizing the rapid and efficient optimization of the vehicle's driving range and energy consumption.

[0168] This embodiment provides a method for simulating the driving range of new energy vehicles. It breaks down the thermal management system and develops reduced-order models for the air conditioning circuit, motor system water circuit, battery water circuit, and passenger compartment, thereby improving the engineering feasibility and model accuracy of the thermal management system.

[0169] This embodiment provides a method for simulating the driving range of new energy vehicles. It decomposes the thermal management system and develops reduced-order sub-models such as the air conditioning circuit, electric drive water circuit, battery water circuit, and passenger compartment, thereby promoting the engineering application value of reduced-order models in the field of vehicle thermal management simulation.

[0170] This embodiment provides a method for simulating the driving range of new energy vehicles. It integrates the developed thermal management system subsystem reduced-order model into the dynamic simulation model and establishes an integrated visualization simulation platform, thereby realizing efficient simulation of the driving range of new energy vehicles at high and low temperatures.

[0171] This embodiment provides a method for simulating the driving range of new energy vehicles. Based on the established thermal management simulation reduced-order model database, it promotes the feasibility of expanding the application of reduced-order models to vehicles on the same platform.

[0172] Example 2

[0173] This embodiment provides a simulation device for the driving range of new energy vehicles.

[0174] A new energy vehicle range simulation device includes:

[0175] The reduced-order sub-model generation module is configured to: decompose the vehicle thermal management system model into an air conditioning system sub-model, a motor system cooling circuit sub-model, a battery system cooling circuit sub-model, and a passenger compartment model; generate a sequence of sample input variables for each sub-model obtained from the decomposition; predict sample output variables for the sample input variable sequence through simulation; map the sample input variable sequence and the sample output variables one-to-one to obtain a training dataset; and train the neural network using the training dataset to obtain the reduced-order sub-model.

[0176] The data acquisition module is configured to acquire the parameters of the controlled component and the vehicle parameters corresponding to multiple consecutive time points as a sequence of input variables;

[0177] The prediction module is configured to: predict the output variables corresponding to the multiple reduced-order sub-models based on the input variable sequence; obtain the load power and system state parameters based on the output variables corresponding to the multiple reduced-order sub-models; calculate the vehicle parameters based on the load power using the vehicle dynamics model; and calculate the parameters of the controlled component based on the system state parameters using a management and control strategy.

[0178] The judgment module is configured to: calculate the driving range when the battery is depleted; and update the input variable sequence using the calculated parameters of the controlled component and the vehicle parameters as the latest data when the battery is not depleted, and update the output variable according to the updated input variable sequence.

[0179] In some embodiments, the input variables of the air conditioning system sub-model include: compressor speed, condenser inlet air temperature, condenser inlet air volume, evaporator inlet air humidity, evaporator inlet air volume, evaporator inlet air temperature, electronic expansion valve superheat, thermal expansion valve superheat, inlet water temperature and inlet water flow of the open-loop battery circuit inlet plate heat exchanger; the output variables of the air conditioning system sub-model include: compressor power, compressor exhaust temperature, blower power, fan power, system refrigerant high pressure, system refrigerant low pressure, HVAC outlet air temperature, plate heat exchanger refrigerant outlet temperature and outlet pressure, and battery inlet water temperature.

[0180] In some embodiments, the input variables of the motor system cooling circuit sub-model include: water pump speed, radiator air intake speed, radiator air intake temperature, motor speed, motor torque, and electronic control heat; the output variables of the motor system cooling circuit sub-model include: coolant flow rate, radiator inlet water temperature, motor controller inlet water temperature, motor body temperature, and electrical power of each component of the motor system.

[0181] In some embodiments, the input variables of the battery system cooling circuit sub-model include: battery inlet water temperature, electric water pump speed, battery charge, current, voltage, and initial battery temperature; the output variables of the battery system cooling circuit sub-model include: coolant flow rate, battery inlet / outlet water temperature, maximum cell temperature, minimum cell temperature, and battery heat generation.

[0182] In some embodiments, the input variables of the occupant cabin model include: air temperature and air volume at the air conditioning unit outlet, ambient temperature, solar radiation intensity, vehicle speed, and initial cabin air temperature; the output variables of the occupant cabin model include: average cabin air temperature and blower power.

[0183] In some embodiments, the reduced-order sub-model generation module is configured as follows:

[0184] Initialize a population containing multiple individuals, each individual corresponding to a neural network and learnable parameters; the neural network corresponds to a number of hidden layers.

[0185] For each individual's corresponding neural network, the training dataset is used for training, the learnable parameters are optimized, and the error value is obtained after training is completed.

[0186] Fitness is determined based on the error value, and screening and genetic operations are performed on the population based on the fitness. The screening and genetic operations are used to determine the optimal individual.

[0187] If the maximum number of iterations is reached, the neural network trained based on the optimal individual is used as the reduced-order sub-model; if the maximum number of iterations is not reached, the number of iterations is incremented by one, and the process returns to train the neural network corresponding to each individual.

[0188] In some embodiments, for the reduced-order sub-model, the number of time points in the input variable sequence used for output variable prediction is consistent with the number of hidden layers corresponding to the optimal individual.

[0189] It should be noted that the new energy vehicle range simulation device provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the electronic device can be divided into different functional modules to complete all or part of the functions described above. In addition, the new energy vehicle range simulation device and the new energy vehicle range simulation method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process is detailed in the method embodiments, which will not be repeated here.

[0190] Example 3

[0191] This embodiment provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in the new energy vehicle range simulation method described in Embodiment 1 above.

[0192] Example 4

[0193] This embodiment provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the new energy vehicle range simulation method described in Embodiment 1 above.

[0194] Figure 5 shows a schematic diagram of the structure of a computer device provided in an exemplary embodiment of this application. The computer device includes a processor 510 and a memory 520.

[0195] The processor 510 may include one or more processing cores, such as a quad-core processor, an octa-core processor, etc. The processor may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field Programmable Gate Array), and PLA (Programmable Logic Array). The processor may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content required to be displayed on the screen. In some embodiments, the processor may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0196] The memory 520 may include one or more computer-readable storage media, which may be non-transitory. The memory may also include high-speed random access memory and non-volatile memory, such as one or more disk storage devices or flash memory devices. In some embodiments, the non-transitory computer-readable storage media in the memory are used to store at least one computer program, which is configured by a processor to implement the in-vehicle occupant detection method provided in the method embodiments of this application.

[0197] Those skilled in the art will understand that the structure shown does not constitute a limitation on the computer device and may include more or fewer components than shown, or combine certain components, or employ different component arrangements.

[0198] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A method for simulating the driving range of a new energy vehicle, the method comprising: Obtain the parameters of the controlled component and the vehicle parameters corresponding to multiple consecutive time points as a sequence of input variables; Based on the input variable sequence, the output variables corresponding to the multiple reduced-order sub-models are predicted through multiple reduced-order sub-models; based on the output variables corresponding to the multiple reduced-order sub-models, the load power and system state parameters are obtained; based on the load power, the vehicle parameters are calculated through the vehicle dynamics model; based on the system state parameters, the parameters of the controlled components are calculated through the management and control strategy. The driving range is calculated when the battery is completely depleted; If the battery power is not depleted, the calculated parameters of the controlled component and the vehicle parameters are used as the latest data to update the input variable sequence, and the output variable is updated according to the updated input variable sequence. The training process of the reduced-order sub-model includes: splitting the vehicle thermal management system model into an air conditioning system sub-model, a motor system cooling circuit sub-model, a battery system cooling circuit sub-model, and a passenger compartment model; generating a sequence of sample input variables for each sub-model; predicting sample output variables for the sequence of sample input variables through simulation; matching the sequence of sample input variables with the sample output variables to obtain a training dataset; and training the neural network using the training dataset to obtain the reduced-order sub-model.

2. The method according to claim 1, wherein, The input variables of the air conditioning system sub-model include: compressor speed, condenser inlet air temperature, condenser inlet air volume, evaporator inlet air humidity, evaporator inlet air volume, evaporator inlet air temperature, electronic expansion valve superheat, thermal expansion valve superheat, and inlet water temperature and flow rate of the inlet plate heat exchanger of the open-loop battery circuit. The output variables of the air conditioning system sub-model include: compressor power, compressor exhaust temperature, blower power, fan power, system refrigerant high pressure, system refrigerant low pressure, HVAC outlet air temperature, plate heat exchanger refrigerant outlet temperature and outlet pressure, and battery inlet water temperature.

3. The method according to claim 1 or 2, wherein, The input variables of the motor system cooling circuit sub-model include: water pump speed, radiator air intake speed, radiator air intake temperature, motor speed, motor torque, and electronic control heat. The output variables of the motor system cooling circuit sub-model include: coolant flow rate, radiator inlet water temperature, motor controller inlet water temperature, motor body temperature, and electrical power of each component of the motor system.

4. The method according to any one of claims 1 to 3, wherein, The input variables of the battery system cooling circuit sub-model include: battery inlet water temperature, electric water pump speed, battery charge, current, voltage, and battery initial temperature. The output variables of the battery system cooling circuit sub-model include: coolant flow rate, battery inlet / outlet water temperature, maximum cell temperature, minimum cell temperature, and battery heat generation.

5. The method according to any one of claims 1 to 4, wherein, The input variables for the passenger compartment model include: air temperature and air volume at the air conditioning unit outlet, ambient temperature, solar radiation intensity, vehicle speed, and initial cabin air temperature. The output variables of the occupant cabin model include: the average temperature of the cabin air and the power of the blower.

6. The method according to any one of claims 1 to 5, wherein, The step of training the neural network using a training dataset includes: Initialize a population containing multiple individuals, each individual corresponding to a neural network and learnable parameters; the neural network corresponds to a number of hidden layers. For each individual's corresponding neural network, the training dataset is used for training, the learnable parameters are optimized, and the error value is obtained after training is completed. Fitness is determined based on the error value, and screening and genetic operations are performed on the population based on the fitness. The screening and genetic operations are used to determine the optimal individual. If the maximum number of iterations is reached, the neural network trained based on the optimal individual is used as the reduced-order sub-model; if the maximum number of iterations is not reached, the number of iterations is incremented by one, and the process returns to train the neural network corresponding to each individual.

7. The method according to claim 6, wherein, For the reduced-order sub-model, the number of time points in the input variable sequence used for output variable prediction is consistent with the number of hidden layers corresponding to the optimal individual.

8. A new energy vehicle range simulation device, comprising: The reduced-order sub-model generation module is configured to: decompose the vehicle thermal management system model into an air conditioning system sub-model, a motor system cooling circuit sub-model, a battery system cooling circuit sub-model, and a passenger compartment model; generate a sequence of sample input variables for each sub-model obtained from the decomposition; predict sample output variables for the sample input variable sequence through simulation; map the sample input variable sequence and the sample output variables one-to-one to obtain a training dataset; and train the neural network using the training dataset to obtain the reduced-order sub-model. The data acquisition module is configured to acquire the parameters of the controlled component and the vehicle parameters corresponding to multiple consecutive time points as a sequence of input variables; The prediction module is configured to: predict the output variables corresponding to the multiple reduced-order sub-models based on the input variable sequence; obtain the load power and system state parameters based on the output variables corresponding to the multiple reduced-order sub-models; calculate the vehicle parameters based on the load power using the vehicle dynamics model; and calculate the parameters of the controlled component based on the system state parameters using a management and control strategy. The judgment module is configured to calculate the remaining driving range when the battery is depleted. If the battery power is not depleted, the calculated parameters of the controlled component and the vehicle parameters are used as the latest data to update the input variable sequence, and the output variable is updated according to the updated input variable sequence.

9. A computer-readable storage medium storing a computer program, wherein the computer program, when executed by a processor, implements the steps in the new energy vehicle range simulation method as described in any one of claims 1 to 7.

10. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in the new energy vehicle range simulation method as described in any one of claims 1 to 7.

Images