Hybrid energy storage system energy management strategy based on adaptive weight and related device

Abstract

The invention belongs to the technical field of energy storage, and particularly relates to a hybrid energy storage system energy management strategy based on adaptive weight and a related device. The invention provides a self-adaptive weight model based on a multi-layer perceptron by adopting a model prediction control method, and the self-adaptive weight model is used for real-time control of a hybrid energy storage system. Specifically, firstly, terminal constraint is applied to the state of charge of the electric energy storage equipment in an optimization function, then optimal weights corresponding to different initial state variables are calculated off line, and the multi-layer perceptron is trained according to the optimal weights. After training is completed, the multi-layer perceptron calculates the weight of the model prediction control module in real time based on the current state variable. And finally, the model predictive control module outputs an input variable of the system as a control reference value based on the obtained weight so as to control the system to operate. Compared with a traditional fixed weight model prediction control method, the weight can be adaptively adjusted according to the real-time state of the hybrid energy storage system, so that the control performance of the hybrid energy storage system is effectively improved.

Description

Technical Field

[0001] The present invention relates to the field of energy storage technology, and in particular to an energy management strategy and related devices of a hybrid energy storage system based on adaptive weights. Background Art

[0002] Energy storage related fields urgently need to explore new energy fuels to reduce dependence on traditional fossil fuels. Hydrogen energy is becoming increasingly important as a clean, pollution-free and widely available new energy source. Fuel cells (FC) have become the main form of hydrogen energy application and have attracted much attention due to their high energy density and clean operation. However, due to the poor dynamic response of fuel cells, it is difficult to meet the instantaneous power demand alone, so hybrid energy storage systems (HESS) that combine fuel cells with electrical energy storage devices have emerged. Among them, hybrid energy storage systems are applied to vehicles and / or renewable energy systems. Vehicles include but are not limited to passenger cars, trucks, buses, ships, and airplanes. Renewable energy systems include but are not limited to power grids and microgrids. Hybrid energy storage systems include at least electrical energy storage devices and fuel cells. Electrical energy storage devices include but are not limited to supercapacitors, lithium-ion batteries, and sodium-ion batteries. Fuel cells include but are not limited to proton exchange membrane fuel cells, alkaline fuel cells, phosphoric acid fuel cells, solid oxide fuel cells, and molten carbonate fuel cells.

[0003] Model Predictive Control (MPC) is an optimization-based method that can be applied to multivariable systems, such as hybrid energy storage systems. However, the traditional MPC method uses fixed weights, so when the state of the hybrid energy storage system changes, it cannot be adjusted in time and effectively, affecting its control performance. Therefore, it is necessary to provide an energy management strategy and related devices for hybrid energy storage systems based on adaptive weights. Summary of the invention

[0004] The present invention provides an energy management strategy and related devices for a hybrid energy storage system based on adaptive weights, which improves the problem in the prior art that energy cannot be flexibly allocated according to the working conditions of the hybrid energy storage system due to fixed MPC weights.

[0005] The present invention provides an energy management strategy for a hybrid energy storage system based on adaptive weights, which is applied to a hybrid energy storage system, and the energy management strategy includes: obtaining the system state of the hybrid energy storage system at the current sampling moment; wherein the system state includes the charge state of the electric energy storage device and the output current of the fuel cell; according to the system state at the current sampling moment, predicting the system state of the hybrid energy storage system at multiple predicted sampling moments based on the state space equation; inputting the system state at the current sampling moment and the system state at multiple predicted sampling moments into a weight prediction model to generate adaptive weights of the hybrid energy storage system at each sampling moment; at each sampling moment, determining the control input of the hybrid energy storage system according to the corresponding adaptive weight and system state; wherein the control input is the output current change of the fuel cell; adjusting the duty cycle of the boost converter in the hybrid energy storage system according to the control input at each sampling moment, and allocating the energy of the fuel cell and the electric energy storage device at the corresponding sampling moment based on the adjusted duty cycle.

[0006] In one embodiment of the present invention, the state space equation is: k+1 =Ax k +B u u k +B d d k , where A is the state coefficient matrix, B u is the input coefficient matrix, B d is the perturbation coefficient matrix, x k is the system state at sampling time k, d k is the load current of the energy storage control system at sampling time k, u k is the control input at sampling time k.

[0007] In one embodiment of the present invention, the system state at the current sampling moment and the system states at multiple predicted sampling moments are input into a weight prediction model to generate the adaptive weights of the hybrid energy storage system at each sampling moment, including: normalizing each system state; inputting each normalized system state into the weight prediction model to predict the adaptive weights of the hybrid energy storage system at each sampling moment.

[0008] In one embodiment of the present invention, the weight prediction model is trained by pre-acquired system state set and corresponding adaptive weight label set. For each system state, the generation process of the corresponding adaptive weight label includes: randomly generating a preset number of adaptive weights; based on a preset cost function, calculating the target cost corresponding to each adaptive weight; iterating and updating each adaptive weight based on the gray wolf optimization algorithm according to the target cost, and when the change in the target cost corresponding to one of the adaptive weights is less than the preset change threshold, it is used as the adaptive weight label corresponding to the system state.

[0009] In one embodiment of the present invention, the adaptive weight includes an adaptive state weight and an adaptive input weight. The control input of the hybrid energy storage system is determined according to the corresponding adaptive weight and the system state at each sampling moment, including: for each sampling moment, a state weight matrix is ​​generated based on the corresponding adaptive state weight; for each sampling moment, an input weight matrix is ​​generated based on the corresponding adaptive input weight; according to a preset control input cost function, the target cost of the control input at all sampling moments is obtained based on the system state, the state weight matrix and the input weight matrix corresponding to all sampling moments; and the control input of the hybrid energy storage system at each sampling moment is obtained by minimizing the target cost.

[0010] In one embodiment of the present invention, the control input of the hybrid energy storage system is determined according to the corresponding adaptive weight and system state at each sampling moment, including: for each sampling moment, generating a state weight matrix based on the corresponding adaptive state weight; for each sampling moment, generating an input weight matrix based on the corresponding adaptive input weight; generating a symmetric weight matrix based on the number of all sampling moments; according to a preset control input cost function, based on the system state corresponding to all sampling moments, the state weight matrix, the input weight matrix and the symmetric weight matrix, obtaining the target cost of the control input at all sampling moments; minimizing the target cost to obtain the control input of the hybrid energy storage system at each sampling moment.

[0011] In one embodiment of the present invention, for each sampling moment, the energy of the hybrid energy storage system at the corresponding sampling moment is distributed according to the control input at the sampling moment, including: based on a proportional-integral algorithm, obtaining the duty cycle of the boost converter of the hybrid energy storage system at the sampling moment according to the control input at the sampling moment; and regulating the energy distribution of the fuel cell and the electrical energy storage device at the corresponding sampling moment based on the duty cycle.

[0012] In one embodiment of the present invention, there is also provided an energy distribution device for a hybrid energy storage system based on adaptive weights, which is applied to a hybrid energy storage system, wherein the energy management device comprises: an initial state acquisition module, which is used to acquire the system state of the hybrid energy storage system at the current sampling moment; wherein the system state comprises the state of charge of the electric energy storage device and the output current of the fuel cell; a state prediction module, which is used to predict the system state of the hybrid energy storage system at multiple predicted sampling moments based on the state space equation according to the system state at the current sampling moment; a weight prediction module, which is used to input the system state at the current sampling moment and the system state at multiple predicted sampling moments into a weight prediction model to generate adaptive weights of the hybrid energy storage system at each sampling moment; an input generation module, which is used to determine the control input of the hybrid energy storage system according to the corresponding adaptive weight and system state at each sampling moment; wherein the control input is the output current change of the fuel cell; an energy management module, which is used to adjust the duty cycle of the boost converter in the hybrid energy storage system according to the control input at each sampling moment, and distribute the energy of the fuel cell and the electric energy storage device at the corresponding sampling moment based on the adjusted duty cycle.

[0013] In one embodiment of the present invention, an electronic device is also provided, comprising: one or more processors; a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device implements any of the above-mentioned energy management strategies for hybrid energy storage systems based on adaptive weights.

[0014] In one embodiment of the present invention, a computer-readable storage medium is further provided, on which a computer program is stored. When the computer program is executed by a processor of a computer, the computer executes any of the above-mentioned energy management strategies for hybrid energy storage systems based on adaptive weights.

[0015] As described above, the energy management strategy and related devices for a hybrid energy storage system based on adaptive weights proposed in the present invention have the following beneficial effects: by collecting the system state of the hybrid energy storage system at the current sampling moment, and predicting the system state at multiple future sampling moments based on the state space equation. The current state and the predicted future state are input into the weight prediction model, and the adaptive weights for each sampling moment are dynamically generated, so that the control process can be adjusted according to the real-time system state. The output current change of the fuel cell is determined according to the adaptive weight and the system state, so as to calculate the control input more accurately and quickly, and optimize the energy distribution of the energy storage system according to the control input. The present invention can adaptively generate dynamic weights for each sampling moment according to the system state, and accurately calculate the control input according to the dynamic weight. It improves the problem that the prior art cannot flexibly allocate energy according to the working conditions of the hybrid energy storage system due to the fixed MPC weight. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of the structure of a fuel cell hybrid vehicle provided by an embodiment of the present invention;

[0017] Figure 2 A schematic diagram showing a flow chart of an energy management strategy for a hybrid energy storage system based on adaptive weights provided by an embodiment of the present invention;

[0018] Figure 3 Shown is a schematic diagram of a process flow of adaptive weight generation provided by an embodiment of the present invention;

[0019] Figure 4 Shown is a structural block diagram of an energy distribution device for a hybrid energy storage system based on adaptive weights provided by an embodiment of the present invention;

[0020] Figure 5 Shown is a structural schematic diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0022] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0023] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.

[0024] The present invention belongs to the field of energy storage technology, and specifically to the field of hybrid energy storage systems composed of fuel cells and electric energy storage devices. The present invention discloses an energy management strategy and related devices for hybrid energy storage systems based on adaptive weights. The present invention adopts a model predictive control method and proposes an adaptive weight model based on a multilayer perceptron for real-time control of hybrid energy storage systems. Specifically, when the hybrid energy storage system is controlled in real time, terminal constraints are first imposed on the state of charge of the electric energy storage device in the optimization function, and then the optimal weights corresponding to different initial state variables are calculated offline, and the multilayer perceptron is trained using the calculated optimal weights. After the training is completed, the multilayer perceptron calculates the weights of the model predictive control module in real time based on the current state variables. Finally, the model predictive control module outputs the input variables of the system as control reference values ​​based on the obtained weights, thereby adjusting the duty cycle of the converter and realizing the control of the hybrid energy storage system. Compared with the traditional fixed-weight model predictive control method, the present invention can adaptively adjust the weights according to the real-time state of the hybrid energy storage system, thereby effectively improving the control performance of the hybrid energy storage system.

[0025] The energy management strategy described in the present invention is applicable to any hybrid energy storage system including a fuel cell, an electric energy storage device, and a boost converter. For ease of description, the present invention takes a hybrid energy storage system including a semi-automatic fuel cell and a supercapacitor in the context of a fuel cell hybrid electric vehicle power system as an example. Figure 1 As shown in the figure, the fuel cell is connected in series with the boost converter as the main power source, the supercapacitor is connected in parallel with the boost converter as an energy buffer, and the load includes components such as DC / AC inverter, motor and transmission. In this topology, L fc Simulating the input inductance of the boost converter, R fc Indicates L fc The parasitic resistance, D fc is the duty cycle signal generated by the PI controller. The system state at each sampling moment is input into the weight prediction model, the corresponding adaptive weight is dynamically generated according to the system state, and the current change of the fuel cell at the corresponding sampling moment is obtained according to the preset cost function. Thus, the duty cycle D of the boost converter is adjusted by the current change. fc , in order to more accurately control the output current i of the fuel cell at each sampling moment fc , thereby changing the output current i of the supercapacitor at the sampling moment sc , to achieve energy distribution between supercapacitor and fuel cell. The output current i of the fuel cell after final distribution is sc The output current i of the supercapacitor sc After merging, a load current i that meets the load requirements is formed dc .

[0026] See also Figure 2 The present invention provides an energy management strategy for a hybrid energy storage system based on adaptive weights. The energy management strategy or the hybrid energy storage system in other embodiments can be applied to a variety of scenarios, including but not limited to energy storage scenarios such as vehicles, renewable energy systems, vehicles including but not limited to passenger cars, trucks, buses, ships, and airplanes, renewable energy systems including but not limited to power grids and microgrids, and hybrid energy storage systems including at least electrical energy storage devices and fuel cells, wherein the electrical energy storage devices include but are not limited to supercapacitors, lithium-ion batteries, and sodium-ion batteries, and the fuel cells include but are not limited to proton exchange membrane fuel cells, alkaline fuel cells, phosphoric acid fuel cells, solid oxide fuel cells, and molten carbonate fuel cells. The energy management strategy for a hybrid energy storage system based on adaptive weights includes the following steps:

[0027] S1. Obtaining the system state of the hybrid energy storage system at the current sampling time; wherein the system state includes the charge state of the electrical energy storage device and the output current of the fuel cell.

[0028] The hybrid energy storage system includes at least a fuel cell, an electrical energy storage device and a boost converter, wherein the fuel cell includes but is not limited to a proton exchange membrane fuel cell, an alkaline fuel cell, a phosphoric acid fuel cell, a solid oxide fuel cell and a molten carbonate fuel cell, etc., and the electrical energy storage device includes but is not limited to a supercapacitor, a lithium-ion battery and a sodium-ion battery, etc. According to the preset sampling interval, the state of charge SOC(k) of the electrical energy storage device at the current sampling time k and the output current i of the fuel cell at the current sampling time k are continuously collected. fc (k), thus obtaining the system state x at the current sampling time k k , and record it as x k =[SOC(k),i fc (k)] T .

[0029] S2. According to the system state at the current sampling moment, the system state of the hybrid energy storage system at multiple predicted sampling moments is predicted based on the state space equation.

[0030] According to the system state x at the current sampling time k k , using the state space equation shown in formula (1), the system state of the hybrid energy storage system at multiple future sampling moments is calculated:

[0031] x k+1 =Ax k +B u u k +B d d k (1)

[0032] Among them, A is the state coefficient matrix, B u is the input coefficient matrix, B d is the perturbation coefficient matrix, x k is the system state at sampling time k, d k is the load current at sampling time k, u k is the control input at sampling time k, x k+1 is the system state at sampling time k+1. State coefficient matrix A, input coefficient matrix B u and the perturbation coefficient matrix B d The calculation method of is shown in formulas (2) to (4):

[0033]

[0034] Among them, η boost is the preset efficiency of the boost converter, v fc (k) is the output voltage of the fuel cell at sampling time k, T s is the sampling time period, C sc is the capacitance of the supercapacitor, v sc (k) is the voltage of the supercapacitor at sampling time k, v max is the maximum voltage of the supercapacitor. The parameters in the above three matrices can all be considered as known quantities.

[0035] S3. Inputting the system state at the current sampling moment and the system states at multiple predicted sampling moments into a weight prediction model to generate adaptive weights of the hybrid energy storage system at each sampling moment.

[0036] The system state sequence {x k+1 ,x k+2 ,…,x k+N}. The system state x at the current sampling time k is k The system states at multiple sampling moments in the future are input into the pre-trained weight prediction model, which generates adaptive weights at the corresponding sampling moments by extracting the state characteristics of the system. The weight prediction model can be any model that can be suitable for weight mapping, including but not limited to Multi-Layer Perceptron (MLP), random forest, convolutional neural network, etc.

[0037] In one embodiment of the present invention, the system state at the current sampling moment and the system states at multiple predicted sampling moments are input into a weight prediction model to generate the adaptive weights of the hybrid energy storage system at each sampling moment, including:

[0038] Normalize the status of each system;

[0039] The normalized system states are input into the weight prediction model to predict the adaptive weight of the hybrid energy storage system at each sampling moment.

[0040] The following processing is performed for each system state that needs to be input into the weight prediction model: the state of charge of the energy storage device and the output current of the fuel cell in the system state are normalized respectively to unify the data to the range acceptable to the model, thereby eliminating the dimensional difference between the data. The normalized system state is input into the weight prediction model, and the adaptive weight of the system state at the sampling moment is predicted by extracting the characteristics of the system state. Among them, the adaptive weight includes the adaptive state weight Q and the adaptive input weight R. The adaptive state weight is used to characterize the importance of the system state deviation, and the adaptive input weight is used to characterize the importance of the control input to the current optimization target. Both of these adaptive weights are symmetric positive definite matrices.

[0041] In one embodiment of the present invention, the weight prediction model is obtained by training a pre-acquired system state set and a corresponding adaptive weight label set. For each system state, the generation process of the corresponding adaptive weight label includes:

[0042] Randomly generate a preset number of adaptive weights;

[0043] Based on the preset cost function, calculate the target cost corresponding to each adaptive weight;

[0044] The adaptive weights are iterated and updated based on the gray wolf optimization algorithm according to the target cost, and when the change of the target cost corresponding to one of the adaptive weights is less than the preset change threshold, it is used as the adaptive weight label corresponding to the system state.

[0045] In the offline stage, it is necessary to generate adaptive weight labels for the system state for training the weight prediction model. Specifically, the Grey Wolf Optimization (GWO) algorithm is used to select the optimal weight group for each system state (i.e., the charge state of the supercapacitor and the output current of the fuel cell) as a training sample. The specific process is as follows: within the preset weight range, a number (e.g., n) of adaptive weight groups are randomly generated, each of which includes an adaptive state weight {q 1,k ,q 2,k} and adaptive input weight r k. The randomly generated n weight groups are used as the initial position of the gray wolf population, and each gray wolf in the population represents a candidate adaptive weight group. The initially generated n adaptive weight groups are substituted into the preset cost function, and the cost values ​​corresponding to each adaptive weight group can be calculated through the system state. The first three adaptive weight groups with the smallest cost values ​​are selected, corresponding to the optimal weight, second-best weight and third-best weight of the current population. The remaining weight values ​​are updated according to these three weights to generate new candidate weight groups. The updated weight groups are used again to calculate the target cost using the cost function. Repeat the above process until the change in the target cost value is less than the preset threshold, and the corresponding optimal weight group is used as the adaptive weight label corresponding to the system state. Through the above gray wolf optimization algorithm, the position of the weight group can be iteratively updated step by step in the randomly generated candidate adaptive weight group with the target cost function as the optimization target, and finally the weight group with the smallest cost function is used as the adaptive weight label of the system state. In this way, more accurate data labels can be provided for the training of subsequent weight prediction models to improve the accuracy of model training.

[0046] S4. At each sampling moment, determine the control input of the hybrid energy storage system according to the corresponding adaptive weight and the system state; wherein the control input is the output current change of the fuel cell.

[0047] At each sampling time k( N is the prediction step length), according to the current system state x k And the adaptive weight obtained by the weight prediction model is input into the cost function, and the change Δi of the output current corresponding to the fuel cell (such as the fuel cell) at the current sampling time k is calculated by minimizing the cost function. fc (k), and use it as the control input u at the current sampling time k k .

[0048] In one embodiment of the present invention, the adaptive weight includes an adaptive state weight and an adaptive input weight, and the control input of the hybrid energy storage system is determined according to the corresponding adaptive weight and the system state at each sampling time, including:

[0049] For each sampling moment: generate a state weight matrix based on the corresponding adaptive state weight;

[0050] For each sampling moment: generate an input weight matrix based on the corresponding adaptive input weight;

[0051] According to the preset cost function of the control input, the target cost of the control input at all sampling times is obtained based on the system state, the state weight matrix and the input weight matrix corresponding to all sampling times;

[0052] The target cost is minimized to obtain the control input of the hybrid energy storage system at each sampling time.

[0053] The system state x at the current sampling time k k , the system state x is obtained through the weight prediction model k The corresponding adaptive weight group, wherein the adaptive weight group includes adaptive state weights and adaptive input weights. The adaptive state weights and adaptive input weights are respectively constructed into the corresponding state weight matrix Q and input weight matrix R as shown in formulas (5) and (6):

[0054]

[0055] R=r k (6)

[0056] Among them, q 1,k ,q 2,k is the adaptive state weight at sampling time k, r k is the adaptive input weight at sampling time k. The system state x at all sampling times k , the state weight matrix Q corresponding to each system state and the input weight matrix R are input into the cost function shown in formula (7):

[0057]

[0058] Among them, x i|k is the system state at future sampling time i predicted at sampling time k, x ref is the reference state, x ref =[SOC ref ,0] T .SOC ref is the charge state reference value of the electric energy storage device (such as supercapacitor), while the current reference value of the fuel cell (such as fuel cell) is set to i fcref = 0. k is the control input sequence, representing all control inputs from the current sampling time k to the end of the prediction, Considering this scheme, x k =[SOC(k),i fc (k)] T ,u k =Δi fc (k) and d k =i load (k). Through optimization algorithms such as quadratic programming or gradient descent, the cost function in (7) is gradually minimized to obtain the optimal control input {Δi fc (k),Δi fc(k+1),Δi fc (k+2),…,Δi fc (k+N-1)}.

[0059] In another embodiment of the present invention, determining the control input of the hybrid energy storage system according to the corresponding adaptive weight and the system state at each sampling moment includes:

[0060] For each sampling moment: generate a state weight matrix based on the corresponding adaptive state weight;

[0061] For each sampling moment: generate an input weight matrix based on the corresponding adaptive input weight;

[0062] Generate a symmetric weight matrix based on the number of all sampling moments;

[0063] According to a preset cost function of the control input, based on the system state corresponding to all sampling moments, the state weight matrix, the input weight matrix and the symmetric weight matrix, the target cost of the control input at all sampling moments is obtained;

[0064] The target cost is minimized to obtain the control input of the hybrid energy storage system at each sampling time.

[0065] For each sampling moment, the system state at the current sampling moment is input into the weight prediction model to obtain the adaptive weight combination at the sampling moment, thereby constructing the state weight matrix And the input weight matrix R = r k According to the total amount N of all sampling moments, a symmetric weight matrix P is constructed. The symmetric weight matrix P is shown in formula (8):

[0066]

[0067] Through the symmetric weight matrix P, the constraints of the future terminal state can be strengthened according to the length of the prediction time domain to minimize the change of the charge state of the electric energy storage device after the drive cycle. Substitute the system state, state weight matrix, input weight matrix and symmetric matrix into the following preset objective function J′(U k )middle:

[0068]

[0069] The optimization algorithm is used to gradually minimize the target cost function, and finally the optimal control input sequence {Δi fc (k),Δi fc (k+1),Δi fc (k+2),…,Δi fc (k+N-1)}.

[0070] S5. Adjust the duty cycle of the boost converter according to the control input at each sampling moment, and distribute the energy of the fuel cell and the electrical energy storage device at the corresponding sampling moment based on the adjusted duty cycle.

[0071] Specifically, in one embodiment of the present invention, for each sampling moment: allocating the energy of the hybrid energy storage system at the corresponding sampling moment according to the control input at the sampling moment includes:

[0072] Based on a proportional-integral algorithm, obtaining a duty cycle of the boost converter of the hybrid energy storage system at the sampling moment according to the control input at the sampling moment;

[0073] The energy distribution between the fuel cell and the electrical energy storage device at corresponding sampling times is regulated based on the duty cycle.

[0074] The duty cycle data D of the boost converter can be obtained at the current sampling time according to the control input through the proportional integral algorithm. fc (t s ) and outputs it as an adjustment signal. By adjusting the duty cycle of the boost converter, the current flowing through the fuel cell and the current flowing through the supercapacitor are changed, thereby realizing power regulation of the hybrid energy storage device at that moment.

[0075] like Figure 3 As shown, in the present invention, in the offline stage (i.e., the training stage of the weight prediction model), firstly, by adding terminal constraints to the cost function, a cost function with terminal constraints is formed, and in a plurality of preset system states, multiple initial adaptive weight groups are randomly initialized for each system state. The Gray Wolf Optimization Algorithm is used to calculate the minimum value of the cost function to obtain an optimal adaptive weight group corresponding to each system state as the adaptive weight label of the system state. The system state is input into a weight prediction model (such as MLP) for training, and the model parameters are updated according to the difference between the predicted adaptive weight and the corresponding label, and finally a trained weight prediction model is obtained. At this time, it can be applied to the real-time stage, and the corresponding adaptive weight can be obtained by inputting the system state at the current moment into the trained weight prediction model, thereby updating the energy distribution of the energy storage system.

[0076] See also Figure 4The hybrid energy storage system in this embodiment or other embodiments can be applied to a variety of scenarios, including but not limited to energy storage scenarios such as vehicles, renewable energy systems, vehicles including but not limited to passenger cars, trucks, buses, ships, and airplanes, renewable energy systems including but not limited to power grids and microgrids, hybrid energy storage systems including at least electrical energy storage devices and fuel cells, electrical energy storage devices including but not limited to supercapacitors, lithium-ion batteries, and sodium-ion batteries, fuel cells including but not limited to proton exchange membrane fuel cells, alkaline fuel cells, phosphoric acid fuel cells, solid oxide fuel cells, and molten carbonate fuel cells. The energy distribution device 100 of the hybrid energy storage system based on adaptive weights includes: an initial state acquisition module 110, a state prediction module 120, a weight prediction module 130, an input generation module 140, and an energy management module 150. The initial state acquisition module 110 is used to obtain the system state of the hybrid energy storage system at the current sampling time; wherein the system state includes the charge state of the electrical energy storage device and the output current of the fuel cell. The state prediction module 120 is used to predict the system state of the hybrid energy storage system at multiple predicted sampling times based on the state space equation according to the system state at the current sampling time. The weight prediction module 130 is used to input the system state at the current sampling moment and the system state at multiple predicted sampling moments into the weight prediction model to generate the adaptive weight of the hybrid energy storage system at each sampling moment. The input generation module 140 is used to determine the control input of the hybrid energy storage system according to the corresponding adaptive weight and system state at each sampling moment; wherein the control input is the output current change of the fuel cell. The energy management module 150 is used to adjust the duty cycle of the boost converter in the hybrid energy storage system according to the control input at each sampling moment, and allocate the energy of the fuel cell and the electrical energy storage device at the corresponding sampling moment based on the adjusted duty cycle.

[0077] For the specific definition of the energy distribution device of the hybrid energy storage system based on adaptive weights, please refer to the definition of the energy management strategy of the hybrid energy storage system based on adaptive weights in the above text, which will not be repeated here. Each module in the above-mentioned energy distribution device of the hybrid energy storage system based on adaptive weights can be implemented in whole or in part by software, hardware and a combination thereof. The above-mentioned modules can be embedded in or independent of the processor in the computer device in hardware format, or can be stored in the memory of the computer device in software format, so that the processor can call the operations corresponding to the above modules.

[0078] It should be noted that, in order to highlight the innovative part of the present invention, the present embodiment does not introduce modules that are not closely related to solving the technical problem proposed by the present invention, but this does not mean that there are no other modules in the present embodiment.

[0079] See also Figure 5The electronic device 1 may include a memory 12, a processor 13 and a bus, and may also include a computer program stored in the memory 12 and executable on the processor 13, such as an energy management program for a hybrid energy storage system based on adaptive weights.

[0080] Among them, the memory 12 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (for example: SD or DX memory, etc.), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 12 can be an internal storage unit of the electronic device 1, such as a mobile hard disk of the electronic device 1. In other embodiments, the memory 12 can also be an external storage device of the electronic device 1, such as a plug-in mobile hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. equipped on the electronic device 1. Further, the memory 12 can also include both an internal storage unit of the electronic device 1 and an external storage device. The memory 12 can not only be used to store application software and various types of data installed in the electronic device 1, such as the code for energy management of a hybrid energy storage system based on adaptive weights, etc., but can also be used to temporarily store data that has been output or is to be output.

[0081] In some embodiments, the processor 13 may be composed of an integrated circuit, for example, a single packaged integrated circuit, or a plurality of packaged integrated circuits with the same or different functions, including one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and combinations of various control chips. The processor 13 is the control core (Control Unit) of the electronic device 1, and uses various interfaces and lines to connect the various components of the entire electronic device 1, and executes or executes programs or modules stored in the memory 12 (such as a hybrid energy storage system energy management program based on adaptive weights, etc.), and calls the data stored in the memory 12 to execute various functions of the electronic device 1 and process data.

[0082] The processor 13 executes the operating system and various installed applications of the electronic device 1. The processor 13 executes the applications to implement the steps in the above-mentioned energy management strategy of the hybrid energy storage system based on adaptive weights.

[0083] Exemplarily, the computer program may be divided into one or more modules, which are stored in the memory 12 and executed by the processor 13 to complete the present application. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, which are used to describe the execution process of the computer program in the electronic device 1. For example, the computer program may be divided into an initial state acquisition module 110, a state prediction module 120, a weight prediction module 130, an input generation module 140, and an energy management module 150.

[0084] The above-mentioned integrated unit implemented in the form of a software function module can be stored in a computer-readable storage medium, and the computer-readable storage medium can be non-volatile or volatile. The above-mentioned software function module is stored in a storage medium, including a number of instructions for enabling a computer device (which can be a personal computer, a computer device, or a network device, etc.) or a processor to perform part of the functions of the energy management strategy of the hybrid energy storage system based on adaptive weights described in various embodiments of the present application.

[0085] In summary, the present invention discloses an energy management strategy and related devices for a hybrid energy storage system based on adaptive weights, which collects the system state of the hybrid energy storage system at the current sampling moment, and predicts the system state at multiple future sampling moments based on the state space equation. The current state and the predicted future state are input into the weight prediction model, and the adaptive weights at each sampling moment are dynamically generated, so that the control process can be adjusted according to the real-time system state. The output current change of the fuel cell is determined according to the adaptive weight and the system state, so as to calculate the control input more accurately and quickly, and optimize the energy distribution of the energy storage system according to the control input. The present invention can adaptively generate dynamic weights at each sampling moment, so as to accurately calculate the control input. It improves the problem that the prior art cannot flexibly allocate energy according to the working conditions of the hybrid energy storage system due to the fixed MPC weights. Therefore, the present invention effectively overcomes the various shortcomings in the prior art and has a high industrial utilization value.

[0086] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. An energy management strategy for a hybrid energy storage system based on adaptive weights, characterized in that: Applied to a hybrid energy storage system, the energy management strategy includes: Acquire the system state of the hybrid energy storage system at the current sampling time; wherein the system state includes the charge state of the electrical energy storage device and the output current of the fuel cell; According to the system state at the current sampling moment, predicting the system state of the hybrid energy storage system at multiple predicted sampling moments based on the state space equation; Inputting the system state at the current sampling moment and the system states at multiple predicted sampling moments into a weight prediction model to generate adaptive weights of the hybrid energy storage system at each sampling moment; At each sampling moment, the control input of the hybrid energy storage system is determined according to the corresponding adaptive weight and the system state; wherein the control input is the output current change of the fuel cell; The duty cycle of the boost converter in the hybrid energy storage system is adjusted according to the control input at each sampling moment, and the energy of the fuel cell and the electrical energy storage device at the corresponding sampling moment is distributed based on the adjusted duty cycle.

2. The energy management strategy of the hybrid energy storage system based on adaptive weight according to claim 1 is characterized in that: The state space equation is: k+1 =Ax k +B u u k +B d d k , where A is the state coefficient matrix, B u is the input coefficient matrix, B d is the perturbation coefficient matrix, x k is the system state at sampling time k, d k is the load current at sampling time k, u k is the control input at sampling time k, x k+1 is the system state at sampling time k+1.

3. The energy management strategy of the hybrid energy storage system based on adaptive weight according to claim 1 is characterized in that: The system state at the current sampling moment and the system states at multiple predicted sampling moments are input into the weight prediction model to generate the adaptive weight of the hybrid energy storage system at each sampling moment, including: Normalize the status of each system; The normalized system states are input into the weight prediction model to predict the adaptive weight of the hybrid energy storage system at each sampling moment.

4. The energy management strategy of the hybrid energy storage system based on adaptive weight according to claim 3 is characterized in that: The weight prediction model is trained by pre-acquired system state set and corresponding adaptive weight label set. For each system state, the generation process of the corresponding adaptive weight label includes: Randomly generate a preset number of adaptive weights; Based on the preset cost function, calculate the target cost corresponding to each adaptive weight; The adaptive weights are iterated and updated based on the gray wolf optimization algorithm according to the target cost, and when the change of the target cost corresponding to one of the adaptive weights is less than the preset change threshold, it is used as the adaptive weight label corresponding to the system state.

5. The energy management strategy of hybrid energy storage system based on adaptive weight according to claim 1, characterized in that: The adaptive weight includes an adaptive state weight and an adaptive input weight. The control input of the hybrid energy storage system is determined according to the corresponding adaptive weight and the system state at each sampling time, including: For each sampling moment, a state weight matrix is ​​generated based on the corresponding adaptive state weight; For each sampling moment, an input weight matrix is ​​generated based on the corresponding adaptive input weight; According to the preset cost function of the control input, the target cost of the control input at all sampling times is obtained based on the system state, the state weight matrix and the input weight matrix corresponding to all sampling times; The target cost is minimized to obtain the control input of the hybrid energy storage system at each sampling time.

6. The energy management strategy of the hybrid energy storage system based on adaptive weight according to claim 1, characterized in that: The adaptive weight includes an adaptive state weight and an adaptive input weight. The control input of the hybrid energy storage system is determined according to the corresponding adaptive weight and the system state at each sampling time, and further includes: For each sampling moment, a state weight matrix is ​​generated based on the corresponding adaptive state weight; For each sampling moment, an input weight matrix is ​​generated based on the corresponding adaptive input weight; Generate a symmetric weight matrix based on the number of all sampling moments; According to a preset cost function of the control input, based on the system state corresponding to all sampling moments, the state weight matrix, the input weight matrix and the symmetric weight matrix, the target cost of the control input at all sampling moments is obtained; The target cost is minimized to obtain the control input of the hybrid energy storage system at each sampling time.

7. The energy management strategy of the hybrid energy storage system based on adaptive weight according to claim 1, characterized in that: For each sampling moment, adjusting the duty cycle of the boost converter in the hybrid energy storage system according to the control input at the sampling moment, and allocating the energy of the fuel cell and the electrical energy storage device at the corresponding sampling moment based on the adjusted duty cycle, including: Based on a proportional-integral algorithm, obtaining a duty cycle of the boost converter of the hybrid energy storage system at the sampling moment according to the control input at the sampling moment; The energy distribution between the fuel cell and the electrical energy storage device at corresponding sampling times is regulated based on the duty cycle.

8. An energy management device for a hybrid energy storage system based on adaptive weights, characterized in that: Applied to a hybrid energy storage system, the energy management device comprises: An initial state acquisition module, used to acquire the system state of the hybrid energy storage system at the current sampling moment; wherein the system state includes the charge state of the electrical energy storage device and the output current of the fuel cell; A state prediction module, used to predict the system state of the hybrid energy storage system at multiple predicted sampling moments based on the state space equation according to the system state at the current sampling moment; A weight prediction module, used to input the system state at the current sampling moment and the system states at multiple predicted sampling moments into a weight prediction model to generate an adaptive weight of the hybrid energy storage system at each sampling moment; An input generation module, used to determine the control input of the hybrid energy storage system according to the corresponding adaptive weight and system state at each sampling moment; wherein the control input is the output current change of the fuel cell; The energy management module is used to adjust the duty cycle of the boost converter in the hybrid energy storage system according to the control input at each sampling moment, and distribute the energy of the fuel cell and the electrical energy storage device at the corresponding sampling moment based on the adjusted duty cycle.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; A storage device for storing one or more programs, which, when executed by the one or more processors, enables the electronic device to implement the energy management strategy of the hybrid energy storage system based on adaptive weights as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and when the computer program is executed by a processor of a computer, the computer is enabled to execute the energy management strategy of the hybrid energy storage system based on adaptive weights as described in any one of claims 1 to 7.

Images