Hybrid energy storage system energy management method based on intelligent network connection technology

Abstract

The invention relates to the technical field of new energy vehicles, and discloses a hybrid energy storage system energy management method based on an intelligent network connection technology, and the method comprises the steps: obtaining the driving state information of a vehicle, and obtaining the external environment information including traffic signal lamp information through intelligent network connection communication; inputting the multi-source information as a time sequence, and predicting a vehicle speed track of the vehicle in a prediction time domain in the future by using a pre-trained long-short-term memory neural network model; establishing an energy management optimization model which takes the minimization of the full-life-cycle comprehensive cost of the hybrid energy storage system as a target function; and carrying out rolling optimization solution on the model by adopting a whale optimization algorithm to obtain an optimal power distribution strategy of the current control period. According to the method, prediction of the future driving working condition is introduced, pre-planning and global optimization of power distribution are achieved, the comprehensive use cost of the energy storage system can be effectively reduced, and the energy utilization efficiency of the whole vehicle is improved.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of new energy vehicles, and particularly relates to an energy management method for a hybrid energy storage system based on intelligent networking technology. BACKGROUND

[0002] With the increasingly severe global energy and environmental problems, new energy vehicles represented by pure electric vehicles have become an inevitable trend of the development of the automobile industry. At the current technical stage, the performance of the vehicle-mounted energy storage system, especially the power battery pack, is one of the core bottlenecks restricting the development of pure electric vehicles. A single type of energy storage element is often difficult to simultaneously meet the requirements of high energy density and high power density. For example, although the widely used lithium ion battery has a relatively high energy density and can ensure the cruising range of the vehicle, its power density is relatively limited, and frequent large-current charging and discharging impacts will seriously affect its service life and safety.

[0003] In order to make up for the performance shortcomings of a single energy storage element, a hybrid energy storage system (HESS) emerges as the times require. The system usually combines an energy-type energy storage element (such as a lithium ion battery) with a power-type energy storage element (such as a super capacitor). In this configuration, the lithium ion battery is responsible for providing the basic and continuous energy required by the vehicle under smooth working conditions, while the super capacitor, with its characteristics of fast charging and discharging and high cycle life, is specially designed to meet the instantaneous high-power demand of the vehicle under acceleration, climbing or braking regeneration and other working conditions.

[0004] However, whether the performance advantages of the hybrid energy storage system can be fully exerted depends largely on the pros and cons of its energy management strategy. Most existing energy management strategies are based on fixed logic thresholds or fuzzy rules. Such strategies only passively allocate power according to the current operating state of the vehicle (such as real-time vehicle speed, demand power, state of charge of the energy storage element, etc.). This reactive control method has inherent limitations because it completely ignores the future driving conditions that the vehicle will face. Due to the lack of foresight, the system cannot make forward-looking global optimization decisions. This often leads to the consumption of super capacitor energy at unnecessary times, while the super capacitor is in poor condition when real high-power impact is needed, forcing the fragile lithium battery to bear excessive current load. Similarly, the system cannot adjust the state in advance to maximize the recovery of braking energy before an explicit deceleration braking event. SUMMARY

[0005] To solve the problems of the prior art, the application provides a hybrid energy storage system energy management method based on intelligent networking technology, which solves the problem that the prior hybrid energy storage system energy management strategy lacks foresight because it only depends on the current driving state, resulting in a locally optimal power distribution decision and the inability to achieve global optimization of the system's full life cycle comprehensive cost and energy efficiency.

[0006] To achieve the above object, the application is implemented by the following technical solutions: the first aspect of the application provides a hybrid energy storage system energy management method based on intelligent networking technology, which is applied to a hybrid energy storage system composed of lithium batteries and supercapacitors, and the method comprises the following steps:

[0007] S1: obtaining driving state information of a driving vehicle, interaction information of the driving vehicle and surrounding vehicles, and roadside infrastructure information;

[0008] S2: based on the obtained driving state information, interaction information and roadside infrastructure information, using a long short-term memory neural network model to predict a predicted future vehicle speed of the driving vehicle within a future preset time period;

[0009] S3: based on the predicted future vehicle speed, establishing a hybrid energy storage system energy management model with the preset cost minimization of the hybrid energy storage system as the optimization target;

[0010] S4: using a whale optimization algorithm to optimize and solve the hybrid energy storage system energy management model to obtain an optimal power distribution strategy of the lithium batteries and the supercapacitors.

[0011] In one specific embodiment, the driving state information includes the driving speed of the current vehicle, the acceleration of the current vehicle, and the distance between the current vehicle and the stop line at the intersection; the interaction information with the surrounding vehicles includes the distance between the current vehicle and the front vehicle, the driving speed of the front vehicle, and the acceleration of the front vehicle; and the roadside infrastructure information includes the state of the traffic signal and the duration of the traffic signal state.

[0012] In one specific embodiment, the step of using a long short-term memory neural network model to predict the predicted future vehicle speed is specifically: forming the driving state information, the interaction information and the roadside infrastructure information collected at past continuous multiple time points into an input feature sequence; inputting the input feature sequence into a pre-trained long short-term memory neural network model to generate the predicted future vehicle speed composed of predicted vehicle speed values at multiple time points within a future preset time period from the model according to the time series dependence relationship contained in the input feature sequence.

[0013] Preferably, the step of establishing the energy management model of the hybrid energy storage system further comprises: calculating the required power of the vehicle in a preset future time period based on a vehicle longitudinal dynamics model according to the predicted future vehicle speed.

[0014] In a specific embodiment, the preset cost is a comprehensive cost of the hybrid energy storage system in a preset life cycle, which includes: a battery pack cost of the lithium battery, a capacitor pack cost of the super capacitor, a DC / DC converter cost, and an electricity cost of the vehicle in the preset life cycle.

[0015] Further, the DC / DC converter cost is determined based on a maximum power absolute value assumed by the super capacitor in the optimal power distribution strategy.

[0016] Further, the electricity cost is calculated based on a unit electricity price, a preset total driving distance of the vehicle in a whole life cycle, and a total electric energy consumed due to internal losses of the lithium battery and the super capacitor in a single driving cycle.

[0017] In a specific embodiment, the step of optimizing and solving the energy management model of the hybrid energy storage system by using the whale optimization algorithm specifically comprises: taking the preset cost as a fitness function for evaluating the pros and cons of the power distribution strategy, and driving the whale optimization algorithm to perform iterative optimization to search for and determine an optimal power distribution strategy composed of the lithium battery power and the super capacitor power, which makes the fitness function value minimum.

[0018] Further, the step of performing iterative optimization by the whale optimization algorithm specifically comprises: initializing a plurality of whale population individuals, wherein a position vector of each individual represents a power distribution strategy; for each iteration, calculating a fitness value corresponding to the power distribution strategy represented by each whale population individual by using the fitness function, and determining a current optimal individual; and updating the position vector of a non-optimal whale population individual to generate a new set of power distribution strategies by mathematical operations simulating the surrounding, spiral updating and searching behaviors of whales, until a preset iteration termination condition is met.

[0019] Preferably, the process of iterative optimization is also limited by the following constraint conditions: the power and state of charge of the lithium battery are within respective preset upper and lower limit ranges; and the power and state of charge of the super capacitor are within respective preset upper and lower limit ranges.

[0020] The present application provides a hybrid energy storage system energy management method based on intelligent networking technology.

[0021] Has the following beneficial effects:

[0022] 1、The application obtains the driving state, surrounding vehicle interaction and roadside facilities and other multi-source information through fusion intelligent network connection technology, and uses a long short-term memory neural network model to predict the future vehicle speed, so that the energy management decision has foresight, and this pre-planning based on future working conditions can avoid the limitations of decisions based on current working conditions, thereby obtaining a more global optimal control effect.

[0023] 2、The energy management model established by the application takes the comprehensive cost including energy storage element loss, power converter hardware specifications and full life cycle electricity consumption as the optimization target, compared with the strategy taking only instantaneous energy consumption as the target, the method can more comprehensively balance the instantaneous performance and long-term economy of the system, and is helpful to reduce the total life cycle cost of the energy storage system.

[0024] 3、The whale optimization algorithm is used to solve the complex energy management model established by the application, which can efficiently search and determine the optimal power distribution strategy in a multidimensional solution space, and ensure the accuracy and realizability of the forward-looking energy management method under the condition of meeting various system operation boundary conditions. BRIEF DESCRIPTION OF DRAWINGS

[0025] Figure 1 It is a hybrid energy storage system energy management method framework of the application;

[0026] Figure 2 It is a schematic diagram of the vehicle at the intersection scene of the application;

[0027] Figure 3 It is a vehicle speed prediction model framework of the application;

[0028] Figure 4 It is a whale optimization algorithm flowchart of the application. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the application will be clearly and completely described below with reference to the drawings in the specification of the application. Obviously, the described embodiments are only part of the embodiments of the application, not all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0030] Referring to the drawings Figure 1 The energy management method provided by the application is executed in a preset control cycle to realize continuous optimization of power distribution of the hybrid energy storage system. In each control cycle, the method mainly includes the following steps.

[0031] Firstly, the method performs an information acquisition step. This step collects data from multiple information sources to build a comprehensive description of the vehicle's current operating state and external environment. Specifically, it not only collects internal driving state information of the vehicle itself, but also obtains external environmental information closely related to the future driving trajectory of the vehicle through intelligent network communication functions, such as the dynamics of the vehicle in front and the state of the roadside facilities at the intersection in front. These data from different dimensions are collected at discrete time points, which together constitute a multi-dimensional data set containing time series features.

[0032] Next, the method performs a future vehicle speed prediction step. This step inputs the multi-dimensional data set containing historical evolution information obtained in the previous step into a pre-trained long short-term memory neural network (LSTM) model as a time series. The model processes the input sequence and generates a new data sequence based on the time sequence dependence learned from the data. The new data sequence is the predicted vehicle speed trajectory of the vehicle in the future preset time period (e.g. 10 seconds). This step completes the key conversion from historical and current state data to future state prediction.

[0033] Subsequently, the method performs an energy management model establishment step. This step aims to convert the physical prediction (future speed trajectory) obtained in the previous step into a solvable mathematical optimization problem. First, based on the longitudinal dynamics model of the vehicle, the predicted speed trajectory is converted into the demand power trajectory of the vehicle at each time point in the future preset time period. Then, around this demand power trajectory, an optimization model is established to minimize the total life cycle cost of the hybrid energy storage system. The model includes an objective function composed of multiple cost items and a series of constraint conditions representing the physical boundaries of the energy storage elements.

[0034] Finally, the method performs a model optimization solving step. This step uses the whale optimization algorithm (WOA) to solve the constrained nonlinear optimization problem established in the previous step. The algorithm iteratively searches the solution space that satisfies all the constraints to find an optimal power distribution trajectory that minimizes the objective function (i.e. the total cost). When the algorithm converges or reaches the preset number of iterations, the optimal power distribution trajectory is obtained. The method extracts the power control command corresponding to the first time step after the current control period from the trajectory and sends it to the bidirectional DC / DC converter in the hybrid energy storage system for execution. After that, in the next control period, the whole process will be repeated again, forming a rolling optimization closed-loop control.

[0035] Referring to the accompanying Figure 2 The information acquisition step of the energy management method of the present application aims to provide the necessary input data for the subsequent future vehicle speed prediction model. This step periodically collects three types of information through on-board sensors and on-board communication modules.

[0036] The first type of information is the driving state information of the vehicle itself. This information is obtained by reading the controller area network (CAN) bus data inside the vehicle. Specifically, it includes: the current vehicle speed calculated from the wheel speed sensor data; the current vehicle acceleration obtained from the inertial measurement unit (IMU) or by differentiating the speed signal; and the distance between the current vehicle and the stop line at the front intersection calculated by comparing the high-precision positioning module (e.g., combined with GPS and RTK technology) with the pre-stored high-precision map data.

[0037] The second type of information is the interaction information with surrounding vehicles. This information is received through the vehicle-mounted vehicle-to-vehicle (V2V) communication module. In a typical intersection scenario, this information mainly refers to the motion information of the vehicle in front of the same lane as the vehicle. Specifically, it includes: the driving speed and acceleration of the vehicle in front obtained by analyzing the V2V communication data packet, and the relative distance to the vehicle in front measured by the radar or camera sensor mounted on the vehicle, or calculated by the V2V positioning information.

[0038] The third type of information is the road-side infrastructure information. This information is received through the vehicle-mounted vehicle-to-infrastructure (V2I) communication module from the road-side unit (RSU) deployed at the intersection. It mainly contains the deterministic information related to the traffic signal, including: the current state of the signal (e.g., red, green or yellow), and the remaining duration of the state. This information provides a direct basis for predicting the passing or stopping behavior of the vehicle at the intersection.

[0039] At each sampling time, the above three types of information are obtained by the vehicle controller, and then subjected to unified timestamp labeling and data alignment processing to ensure the synchronization of all data in the time dimension. Subsequently, these processed data points are combined into a single feature vector. This feature vector completely describes the comprehensive state of the vehicle itself and its micro-traffic environment at that moment. Arranging the feature vectors of continuous multiple sampling times in chronological order forms the input sequence required for the subsequent long short-term memory neural network model.

[0040] The feature vectors of continuous multiple sampling time points are arranged in time sequence, that is, the input sequence required by the subsequent long short-term memory neural network model is formed. Before the input sequence is sent into the neural network model, a data normalization preprocessing step is also included. This step scales each feature value in the input sequence to the interval [-1, 1] through methods such as Min-Max Scaling. This can eliminate the adverse effects of different physical dimensions on model training and speed up the convergence of the model. The maximum and minimum values used for normalization are determined and stored during the model training phase and are called upon during the prediction phase to process real-time collected data. Accordingly, after the model outputs the predicted vehicle speed, a reverse normalization operation is also needed to restore it to a speed value with actual physical meaning.

[0041] Referring to the drawings Figure 3 The future vehicle speed prediction step of the present application aims to process the time series data obtained in the previous step and generate a predicted vehicle speed sequence representing the future driving trend of the vehicle. This step is performed by a pre-trained long short-term memory neural network (LSTM) model.

[0042] This step is performed by a pre-trained long short-term memory neural network (LSTM) model. The pre-trained model is obtained through an offline training process. The process specifically includes: first, collecting a large amount of real vehicle driving data to build a dataset containing multi-source information input sequences and corresponding actual future vehicle speed sequences; then, dividing the dataset into training, validation, and test sets; next, building a neural network model with input layer, LSTM layer, and output layer structure; finally, using Mean Squared Error (MSE) as the loss function, using optimizers such as Adam to iteratively train the model on the training set, and adjusting hyperparameters and preventing overfitting through the performance of the validation set until the model converges, ultimately obtaining a set of fixed model weight parameters that can accurately map from input information to future vehicle speed. The weight parameters are stored in the non-volatile memory of the vehicle controller.

[0043] The input of this step is the input sequence formed by arranging the multi-source information feature vectors obtained at the past N consecutive sampling time points in time sequence. If M features (such as vehicle speed, distance to the vehicle in front, signal light remaining time, etc.) are collected at each sampling time point, the input sequence is an N x M matrix. This structure encapsulates the historical state changes of the vehicle and its environment as a whole for the model to process.

[0044] The long short-term memory neural network model comprises an input layer, one or more LSTM layers, and an output layer. The input layer receives an N x M input sequence. The LSTM layer is the core of processing sequence information, and each LSTM layer is composed of a plurality of neurons with memory units. Each memory unit includes an input gate, a forget gate, and an output gate. These gate structures enable selective updating of the unit state and information transmission to the next time step during the processing of sequence data, thereby effectively capturing long-range temporal dependencies in the input sequence.

[0045] When the input sequence passes through the LSTM layer, the model gradually processes the feature vector of each time step in the sequence according to the weight parameters determined by pre-training, and updates the hidden state. Finally, the last LSTM layer generates a final hidden state vector after processing the entire input sequence. The vector condenses the key information in the entire historical input sequence.

[0046] The final hidden state vector is then sent to the output layer. The output layer is a fully connected layer that linearly transforms the hidden state vector received from the LSTM layer into an output vector of a predetermined length P. The output vector is the final predicted future vehicle speed generated in this step, and the P elements correspond to the predicted vehicle speed values for the next P time steps. For example, if the prediction time domain is 10 seconds and the control period is 1 second, P is 10, and the output is a sequence containing 10 predicted vehicle speed values. This sequence will be directly used as the basis for calculating the future demand power of the vehicle in the next step.

[0047] After completing the future vehicle speed prediction, the method of the present application then performs the establishment step of the hybrid energy storage system energy management model. The purpose of this step is to convert the physical prediction output from the previous step (i.e., the predicted future vehicle speed sequence) into a formalized mathematical problem that can be solved by an optimization algorithm. This step first calculates the future demand power of the vehicle, and then constructs an objective function with the comprehensive cost as the optimization objective based on this. To achieve the above purpose, first, based on the vehicle longitudinal dynamics model, the predicted future vehicle speed sequence v(k) is converted into the demand power P req (k) of the vehicle at each future time step k. The demand power is calculated by the following formula:

[0048]

[0049] where P req (k) is the demand power of the vehicle at time step k; m is the mass of the vehicle; g is the acceleration due to gravity; f is the rolling resistance coefficient; p is the air density; C D is the air resistance coefficient; A is the frontal area of the vehicle; v(k) is the predicted vehicle speed at time step k; and dv(k) / dt is the predicted acceleration at time step k.

[0050] The demand power P req (k) of the vehicle at any time step k bat (k) and the power provided by the supercapacitor

[0051] P sc (k) jointly satisfy, and their relationship follows the power balance equation:

[0052] P req (k) = P bat (k) + P sc (k) ;

[0053] The present application takes the minimization of the comprehensive cost of the hybrid energy storage system within a preset life cycle as the optimization goal. The comprehensive cost objective function J is determined by the following formula:

[0054]

[0055] Wherein, L cyc is the mileage of a single driving cycle, and C bat , C sc , C dcdc and C elec are the lithium battery pack cost, supercapacitor pack cost, DC / DC converter cost and electricity cost respectively.

[0056] The DC / DC converter cost C dcdc is related to the maximum power it needs to bear in operation, and its cost is calculated by the following formula:

[0057] C dcdc = c dcdc · max(|P sc (k)|) ;

[0058] Wherein, c dcdc is the DC / DC converter cost coefficient per unit power; max(|P sc (k)|) is the absolute value of the maximum instantaneous power borne by the supercapacitor within the entire prediction time domain. The lithium ion battery pack cost C bat is determined based on the capacity of its initial configuration, and its cost is calculated by the following formula:

[0059] C bat = c bat · E bat ;

[0060] Wherein, c bat is the lithium battery price per unit energy capacity; E bat is the total energy capacity of the configured lithium battery pack.

[0061] Cost of supercapacitor pack C sc Based on its initial configuration of capacity, its cost is calculated by the following equation:

[0062] C sc = c sc · E sc ;

[0063] where c sc is the price of supercapacitor per unit of energy capacity; E sc is the total energy capacity of the configured supercapacitor pack. The total electricity cost C elec of driving the vehicle in a pre-set life cycle is calculated based on the internal power loss of the system during operation, whose cost is determined by the following equation:

[0064]

[0065] where c e is the unit electricity price; L total is the total driving distance of the vehicle in the pre-set life cycle; T is the total number of steps in the prediction horizon; I bat (k) and I sc (k) are the currents flowing through the lithium battery and supercapacitor at time step k, respectively; R bat and R sc are the equivalent internal resistances of the lithium battery and supercapacitor, respectively; and Δt is the time step length.

[0066] To calculate the above electricity cost, the known power values need to be converted to current values. The currents I bat (k) and I sc (k) can be calculated by the equivalent circuit model of the respective energy storage elements. For example, based on an R int model containing open-circuit voltage and internal resistance, the current can be determined from its power, open-circuit voltage, and internal resistance by the following relationship:

[0067]

[0068] where I(k) represents I bat (k) or I sc (k); P(k) represents the corresponding P bat (k) or P sc (k); U oc (k) is the open-circuit voltage of the energy storage element at the current state of charge; and R int is the corresponding equivalent internal resistance R bat or R sc .

[0069] Reference is made to the accompanying drawings that form a part of this disclosure. Figure 4After the energy management model of the hybrid energy storage system is established, the method of the application optimizes and solves the model by using a whale optimization algorithm (WOA). The purpose of this step is to search for and determine an optimal power distribution strategy that can make the objective function J minimum.

[0070] In the application of the application, each whale individual of the whale optimization algorithm represents a potential solution, and its position vector corresponds to a complete power distribution sequence in the entire prediction time domain. Specifically, the dimension of the position vector is equal to the total number of steps T of the prediction time domain, and each element in the vector represents the super capacitor power value at a time step, that is, The pros and cons of the solution represented by each whale individual are evaluated by substituting the power distribution sequence corresponding to the whale individual into the objective function J established in the previous step, and the value of the objective function J is used as the fitness value of the algorithm.

[0071] The optimization and solution process is initialized first. The running parameters of the algorithm are set, including the size of the whale population and the maximum number of iterations. Then, the initial whale population is randomly generated in the preset solution space, that is, a plurality of random power distribution sequences are generated as the initial solution set.

[0072] Next, the algorithm enters the iteration optimization loop. In each iteration, first, the fitness value corresponding to the position of each whale individual in the population is calculated according to the objective function. Then, the fitness values of all individuals are compared, and the position of the individual with the minimum fitness value in the current population is found and recorded as the current optimal position

[0073] Subsequently, the algorithm updates the positions of other non-optimal individuals in the population by simulating the hunting behavior of humpback whales, so that the positions of the individuals approach the current optimal position. The position updating process includes mechanisms such as surrounding prey, spiral updating (bubble net attack), and random search. According to the technical disclosure document, in a specific implementation, the behavior of surrounding prey can be performed by the following position updating formula:

[0074]

[0075] where t is the current iteration number; is the updated position vector of the individual; is the position vector of the current optimal individual. The vector and are coefficient vectors for controlling position updating, and their calculation method follows the definition of the standard whale optimization algorithm to achieve a balance between exploration and utilization of the algorithm.

[0076] Throughout the whole process of iteration, the solution must always satisfy the physical boundary of the hybrid energy storage system, i.e. the constraint conditions. These constraint conditions are defined by the following inequalities:

[0077] P bat,min ≤P bat (k)≤P bat,max ;

[0078] P sc,min ≤P sc (k)≤P sc,max ;

[0079] SOC bat,min ≤SOC bat (k)≤SOC bat,max ;

[0080] SOC sc,min ≤SOC sc (k)≤SOC sc,max ;

[0081] where P bat,min and P bat,max are the minimum and maximum allowable power of the lithium battery respectively; P sc,min and P sc,max are the minimum and maximum allowable power of the supercapacitor respectively; SOC bat,min and SOC bat,max are the lower and upper limits of the state of charge of the lithium battery respectively; SOC sc,min and SOC sc,max are the lower and upper limits of the state of charge of the supercapacitor respectively. After each new individual position is calculated, it is necessary to check whether the power sequence represented by the position and the whole system state sequence derived therefrom satisfy all the above constraints. For the solutions that exceed the boundary, they need to be adjusted to the boundary value to ensure the physical feasibility of all potential solutions.

[0082] For the solutions that exceed the boundary, they need to be adjusted to the boundary value to ensure the physical feasibility of all potential solutions. In another preferred embodiment, the processing of the constraint conditions is realized by introducing a penalty function mechanism. Specifically, when calculating the fitness value, the original objective function JJ is added with a penalty term to form a new fitness function. When a solution (i.e. a power allocation sequence) causes any constraint condition to be violated, the penalty term is a huge positive value; while when the solution is within the feasible region, the penalty term is zero. In this way, any "whale individual" that violates the constraints will obtain a very poor fitness value, and thus be naturally eliminated in the iterative selection process of the algorithm, guiding the search process to always be within or near the boundary of the feasible region, and finally obtaining the optimal solution that satisfies all the constraints.

[0083] The algorithm iterates until a preset termination condition is met, such as reaching the maximum number of iterations. When the iteration terminates, the algorithm outputs the finally found global optimum position. This vector represents the optimal supercapacitor power allocation sequence for the future prediction time domain. The method of this invention extracts the first element from this optimal sequence. The corresponding lithium battery power was calculated by combining the power balance equation. These two values ​​are used as the final output of the current control cycle and sent down to the underlying controller for execution.

[0084] The energy management method provided by this invention is implemented by a power system installed on a new energy vehicle. At the hardware level, the system mainly consists of a hybrid energy storage system, a drive motor and its controller, and a vehicle controller as the core computing and control unit.

[0085] The hybrid energy storage system is the direct control object of this invention. Its internal topology is specially designed and consists of a lithium-ion battery pack, a supercapacitor pack, and a bidirectional DC / DC converter connected to the supercapacitor branch. This design clarifies the different functions of the two energy storage elements: the lithium-ion battery pack, as the high-energy-density main energy storage unit, is responsible for providing continuous energy during smooth vehicle operation; the supercapacitor pack, as the high-power-density auxiliary power unit, plays a crucial role in responding to drastic fluctuations in vehicle power demand, such as providing instantaneous high-power output during vehicle start-up or rapid acceleration, or quickly and efficiently absorbing the instantaneously generated high-power braking energy during braking.

[0086] In terms of internal electrical connections, the output of the lithium-ion battery pack is directly connected in parallel to the vehicle's high-voltage DC bus. The supercapacitor pack is connected in parallel to the same high-voltage DC bus via the output of the aforementioned bidirectional DC / DC converter. This high-voltage DC bus serves as a power collection and distribution bus, with its other end connected to the drive motor controller to drive the vehicle. This specific connection architecture allows the vehicle controller to actively and precisely adjust the power distribution ratio between the supercapacitor pack and the lithium-ion battery pack by accurately controlling the bidirectional DC / DC converter.

[0087] The vehicle controller is the physical computing platform for the energy management method of this invention, and its internal hardware includes at least one processor and a memory. In one specific embodiment, the processor can be a high-performance microcontroller (MCU) with a floating-point unit, or a digital signal processor (DSP), to meet the needs of real-time calculation of complex mathematical models. The memory may include random access memory (RAM) for temporarily storing computational data, and flash memory or electrically erasable programmable read-only memory (EEPROM) for permanently storing the program code and pre-trained model parameters of this invention.

[0088] The memory pre-stores program code for executing each step of the method of the present invention. The vehicle controller interacts with other sub-controllers of the vehicle chassis and powertrain via the vehicle's Controller Area Network (CAN) bus to periodically acquire real-time driving status information of the vehicle itself. This information includes at least the current vehicle speed obtained by processing data from wheel speed sensors, and the current acceleration obtained by the inertial measurement unit or by differentiating the vehicle speed.

[0089] To predict future operating conditions, the vehicle controller also integrates an onboard communication module supporting vehicle-to-everything (V2X) wireless communication technology. This module has vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication capabilities, enabling it to receive dynamic and static information from the vehicle's external environment. Specifically, this module receives kinematic information (such as speed and acceleration) from vehicles ahead, as well as the current state and remaining duration of traffic lights broadcast by roadside units at the intersection ahead.

[0090] In actual operation, the vehicle controller integrates multi-source data streams from the internal CAN bus and the external V2X communication module, and executes the program code in the processor's memory, which is the energy management method disclosed in this invention. After a series of calculations, including vehicle speed prediction, model building, and optimization, the method finally generates the optimal power allocation instruction for the next control time step. The core of this instruction is to control the operating state of the bidirectional DC / DC converter (such as the chopper duty cycle), thereby precisely controlling the charging and discharging power of the supercapacitor pack, and indirectly determining the power load that the lithium-ion battery pack needs to bear, ultimately realizing online optimization management of the energy flow of the entire hybrid energy storage system.

Claims

1. An energy management method for hybrid energy storage systems based on intelligent network technology, applied to hybrid energy storage systems composed of lithium batteries and supercapacitors, characterized in that, Includes the following steps: Acquire driving status information of the vehicle, interaction information between the vehicle and surrounding vehicles, and roadside infrastructure information; Based on the acquired driving status information, interaction information, and roadside infrastructure information, a long short-term memory neural network model is used to predict the predicted future speed of the vehicle within a preset time period. Based on the predicted future vehicle speed, an energy management model for the hybrid energy storage system is established with the optimization objective of minimizing the preset cost of the hybrid energy storage system. The energy management model of the hybrid energy storage system is optimized and solved using the whale optimization algorithm to obtain the optimal power allocation strategy for the lithium battery and the supercapacitor.

2. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 1, characterized in that, The driving status information includes: the current vehicle speed, the current vehicle acceleration, and the current distance between the vehicle and the stop line at the intersection; The interaction information with surrounding vehicles includes: the distance between the current vehicle and the vehicle in front, the speed of the vehicle in front, and the acceleration of the vehicle in front. The roadside infrastructure information includes: the status of traffic lights and the duration of the traffic light status.

3. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 1, characterized in that, The step of using a long short-term memory neural network model to predict the future speed of the vehicle within a preset time period is as follows: The driving status information, interaction information, and roadside infrastructure information collected at multiple consecutive time points in the past are formed into an input feature sequence; The input feature sequence is input into a pre-trained long short-term memory neural network model, which generates a predicted future vehicle speed consisting of predicted vehicle speed values ​​at multiple time points within a preset future time period, based on the temporal dependencies contained in the input feature sequence.

4. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 1, characterized in that, The step of establishing a hybrid energy storage system energy management model with the preset cost minimization as the optimization objective further includes: Based on the predicted future speed of the vehicle within a preset time period, the power demand of the vehicle within the preset time period is calculated using a vehicle longitudinal dynamics model.

5. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 1, characterized in that, The preset costs include: the cost of the lithium battery pack, the cost of the supercapacitor pack, the cost of the DC / DC converter, and the electricity cost of the vehicle over its preset lifespan.

6. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 5, characterized in that, The cost of the DC / DC converter is determined based on the maximum absolute value of the power that the supercapacitor undertakes in the optimal power allocation strategy.

7. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 5, characterized in that, The electricity cost is calculated based on the unit electricity price, the preset total mileage of the vehicle throughout its entire life cycle, and the total electrical energy consumed in a single driving condition.

8. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 1, characterized in that, The specific steps for optimizing the energy management model of the hybrid energy storage system using the whale optimization algorithm are as follows: The preset cost serves as a fitness function to evaluate the merits of the power allocation strategy and drives the whale optimization algorithm to perform iterative optimization, searching for and determining the optimal power allocation strategy composed of the lithium battery power and the supercapacitor power that minimizes the fitness function value, thereby achieving the optimal energy management model of the hybrid energy storage system.

9. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 8, characterized in that, The steps for determining the optimal power allocation strategy through iterative optimization specifically include: Initialize the locations of multiple whale population individuals representing different power allocation strategies; Based on the fitness function, calculate the fitness value corresponding to the power allocation strategy represented by each individual in the whale population, and determine the current optimal individual position; By simulating the encirclement, spiraling, and searching behaviors of whales, the positions of individual whales in the population are iteratively updated, thereby optimizing and adjusting the power of the lithium battery and the supercapacitor until a preset iteration termination condition is met.

10. The energy management method for a hybrid energy storage system based on intelligent network technology according to claim 8, characterized in that, The iterative optimization process is also subject to the following constraints: The power and state of charge of the lithium battery are within their respective preset upper and lower limits. The power and state of charge of the supercapacitor are within their respective preset upper and lower limits.

Images