An energy-saving optimization method for an electric energy storage device

Abstract

The present invention relates to the technical field of power supply devices, and particularly to an energy-saving optimization method for an electrical energy storage device. By integrating the solid-liquid phase change kinetics equation of a lithium-ion battery, the fluid mechanics equation of a flow battery, and the rigid-flexible coupling equation of a flywheel rotor using neural differential equations, a dynamic meta-knowledge base containing cross-time-scale nonlinear interaction characteristics is constructed; a physics-informed neural network is used to perform dynamic parameter compensation on the polarization voltage hysteresis effect and the electrolyte flow resistance coupling effect to generate a decoupled residual feature vector; combined with a multi-agent reinforcement learning framework, hierarchical control parameters for supercapacitors and damping compensation amounts for flywheel magnetic bearings are generated to form a cooperative control strategy. The present invention realizes multi-physical-field coupling modeling through a dynamic meta-knowledge base, and combines a physics-informed neural network and a quantum optimization algorithm to improve the energy efficiency optimization effect of the energy storage device.

Description

Technical Field

[0001] The present invention relates to the technical field of power supply devices, and particularly to an energy-saving optimization method for an electric energy storage device. Background Art

[0002] Electric energy storage devices store and release electric energy through electrochemical, physical or electromagnetic mechanisms. Common types include lithium-ion batteries, flow batteries, supercapacitors and flywheel energy storage systems. Lithium-ion batteries rely on the redox reaction of lithium ions intercalating and deintercalating between the positive and negative electrodes, and have high energy density and cycle stability; flow batteries use the circulation of electrolytes between external storage tanks and stacks for energy storage, and are suitable for large-scale long-term energy storage scenarios; supercapacitors achieve rapid charge separation based on the electric double layer or pseudocapacitance effect, and have millisecond-level response and millions of cycle characteristics; flywheel energy storage converts electric energy into mechanical energy through a high-speed rotor and uses magnetic levitation technology to reduce friction loss, and is suitable for short-term high-frequency power compensation. Various energy storage devices realize power interaction with the power grid through power electronic converters, and support functions such as frequency regulation, renewable energy consumption and grid-connected / off-grid switching of microgrids.

[0003] The technical pain points affecting the energy-saving control accuracy of electric energy storage devices stem from the synergistic effect of multi-dimensional non-linear dynamic characteristics and system-level coupling interference. In the field of electrochemical energy storage, the dynamic hysteresis effect between the internal polarization voltage and the state of charge of lithium-ion batteries leads to parameter mismatch of the equivalent circuit model, especially causing prediction deviation of the charge and discharge cut-off voltage during frequent working condition switching; flow batteries are affected by the dynamic coupling of the ion migration rate of electrolytes and the pressure drop in the stack flow channels, resulting in lag in the compensation of pump power consumption by the energy management strategy. In power-type energy storage systems, the competitive relationship between the electric double layer charge relaxation effect and the pseudocapacitance reaction kinetics of supercapacitors weakens the linear control accuracy of the terminal voltage in the constant power mode; in flywheel energy storage, due to the interaction between the gyroscopic effect of the high-speed rotor and the non-linear friction force of the bearing, it is difficult to suppress the torque fluctuation during the mechanical energy feedback stage. When the power electronic converter realizes multi-modal switching, the cross-modulation interference between the DC bus voltage ripple and the AC grid-connected harmonic components exacerbates the attenuation of the phase margin of the power closed-loop control. The above dynamic processes and external environmental disturbances form multi-dimensional time-varying constraints, significantly reducing the robustness of the energy efficiency optimization of the energy storage system on a wide time scale (from seconds to hours). Summary of the Invention

[0004] Aiming at the deficiencies of the prior art, the present invention provides an energy-saving optimization method for an electric energy storage device, which solves the problems of model mismatch and control deviation of the energy storage device caused by multi-physical field coupling and non-linear dynamic characteristics.

[0005] To solve the above technical problems, the specific technical solutions of the present invention are as follows:

[0006] The present invention provides an energy-saving optimization method for an electrical energy storage device, including:

[0007] Construct a multi-physics field coupling model of the energy storage device, and fuse the solid-liquid phase change kinetics equation of the lithium-ion battery, the hydrodynamics equation of the flow battery, and the rigid-flexible coupling equation of the flywheel rotor through neural differential equations to generate a dynamic meta-knowledge base containing cross-time-scale non-linear interaction characteristics;

[0008] Based on the physical constraint conditions of the dynamic meta-knowledge base, use a physics-informed neural network to perform dynamic parameter compensation processing on the polarization voltage hysteresis effect and the electrolyte flow resistance coupling effect of the energy storage device, and output a residual feature vector containing the decoupled state-of-charge characteristics;

[0009] Input the residual feature vector into a multi-agent reinforcement learning framework to generate the hierarchical control parameters of the supercapacitor terminal voltage and the magnetic bearing damping compensation amount of the flywheel energy storage, and form a cooperative control strategy;

[0010] According to the power distribution threshold of the cooperative control strategy, use a quantum-inspired optimization algorithm to transform the DC bus ripple suppression and harmonic distortion rate optimization objectives into Hamiltonian constraint conditions of a quantum annealing model, and generate a converter modulation instruction;

[0011] In the federated reinforcement learning framework, distill the cloud global optimization strategy to the edge controller through an imitation learning generative adversarial network, and at the same time inject the real-time operating condition data fed back by the edge controller into the multi-physics field coupling model for parameter recalibration, forming a closed-loop energy efficiency enhancement loop across physical models, data-driven, and quantum optimization.

[0012] Furthermore, for the energy-saving optimization method of the electrical energy storage device of the present invention, the construction of the multi-physics field coupling model of the energy storage device includes:

[0013] Use neural differential equations to fuse the solid-liquid phase change reaction kinetics equation of the lithium-ion battery, the hydrodynamics equation of the flow battery, and the rigid-flexible coupling equation of the flywheel rotor;

[0014] Extract the spatio-temporal correlation characteristics of multi-source heterogeneous data of the energy storage device through a Transformer architecture with spatio-temporal attention mechanism, and generate a meta-knowledge base containing cross-scale non-linear interaction relationships, which is used to provide physical constraint conditions for the physics-informed neural network.

[0015] Furthermore, for the energy-saving optimization method of the electrical energy storage device of the present invention, the dynamic parameter compensation includes:

[0016] On the lithium-ion battery side of the energy storage device, embed the Butler–Volmer electrochemical equation into a long short-term memory network to decouple the cross-sensitivity of the state of charge and the state of health in real time;

[0017] On the side of the flow battery, a graph neural network is constructed to predict the pressure drop distribution in the flow channel. Combining federated learning to aggregate the operation data of multi-node pump units, the feedforward compensation coefficient matrix of the electrolyte flow rate is dynamically optimized, and the compensation residual is used as the state observation input of the multi-agent reinforcement learning.

[0018] Furthermore, in the energy-saving optimization method of the energy storage device of the present invention, the residual feature vector is input into the multi-agent reinforcement learning framework to generate the hierarchical control parameters of the supercapacitor terminal voltage and the magnetic bearing damping compensation amount of the flywheel energy storage, and the collaborative control strategy includes:

[0019] Adopt a hierarchical attention mechanism for the control of the supercapacitor terminal voltage, and dynamically adjust the strategy through sliding mode parameters to match the charge relaxation time constant;

[0020] Deploy a rotor deformation observer based on neural radiance fields on the flywheel energy storage side, and combine the proximal policy optimization algorithm to generate the magnetic bearing damping compensation amount, so that the quantum-inspired optimization algorithm receives the torque fluctuation suppression threshold as the modulation target constraint.

[0021] Furthermore, in the energy-saving optimization method of the energy storage device of the present invention, the multi-objective optimization includes:

[0022] Convert the DC bus ripple suppression, harmonic distortion rate optimization and switching loss minimization into the Hamiltonian of the quantum annealing model;

[0023] Predict the harmonic spectrum distribution characteristics through a deep residual network, and feedback the distribution characteristics to the multi-physical field coupling model to update the boundary conditions of the neural differential equation.

[0024] Furthermore, in the energy-saving optimization method of the energy storage device of the present invention, the formation of a closed-loop energy efficiency enhancement loop of cross-physical models, data-driven and quantum optimization includes:

[0025] Based on the output of the multi-agent reinforcement learning framework, deploy the multi-agent proximal policy optimization algorithm in the cloud, and fuse the operation data of multi-node pump units aggregated by the federated learning through the differential privacy protection mechanism;

[0026] On the edge side, use an imitation learning generative adversarial network to convert the cloud policy into a lightweight control instruction, and inject the instruction execution result back into the generated dynamic meta-knowledge base to trigger the recalibration of the boundary condition parameters of the neural differential equation.

[0027] Furthermore, in the energy-saving optimization method of the energy storage device of the present invention, the federated reinforcement learning framework includes:

[0028] Based on the physical constraints of the dynamic meta-knowledge base, design a hierarchical reward mechanism, map the dynamic efficiency of the energy storage system to the depth of the potential energy well, and construct a long-term energy efficiency optimization goal;

[0029] Generate a dynamic energy efficiency entropy index through a multi-entropy maximization controller, and the index is used to constrain the charge relaxation threshold of the supercapacitor power distribution interval in the generated cooperative control strategy.

[0030] Furthermore, the energy-saving optimization method of the electric energy storage device according to the present invention further includes:

[0031] Embed an adversarial curriculum generator within the federated learning framework to synthesize training scenarios based on the critical operating condition data stream;

[0032] The federated learning framework performs differential privacy processing on the multi-node pump operation data to generate differentially private protected data;

[0033] Evaluate the distribution difference degree between the synthesized scenario and the differentially private protected data through a graph attention network, and dynamically adjust the training curriculum difficulty coefficient of the multi-agent reinforcement learning framework.

[0034] Furthermore, the energy-saving optimization method of the electric energy storage device according to the present invention further includes:

[0035] Construct an unsteady Pareto front search engine, and dynamically allocate multi-objective optimization weights according to the dynamic energy efficiency entropy index and the harmonic spectrum distribution characteristics;

[0036] Input the weight sequence into a quantum-inspired optimization algorithm, and update the Hamiltonian constraint conditions of the quantum annealing model to achieve real-time alignment of the converter modulation strategy and the dynamic energy efficiency optimization target.

[0037] Furthermore, for the energy-saving optimization method of the electric energy storage device according to the present invention, the closed-loop energy efficiency enhancement loop includes:

[0038] Inject the synthesized critical operating condition data stream reversely into the dynamic meta-knowledge base to trigger the reparameterization of the fluid dynamics equation of the flow battery in the neural differential equation;

[0039] Constrain the frequency response interval of the flywheel energy storage magnetic bearing damping compensation amount in the generated cooperative control strategy through a hierarchical reward mechanism and the dynamic energy efficiency entropy index;

[0040] Synchronously feedback the updated dynamic meta-knowledge base parameters to the physical information neural network and the quantum optimization algorithm to form a closed-loop enhancement loop for cross-model parameter compensation, control strategy generation, and modulation optimization.

[0041] Advantages of the present invention;

[0042] The beneficial effects of the present invention are reflected in achieving precise modeling of the multi - physical - field coupling model through a dynamic meta - knowledge base, using a physics - informed neural network to perform real - time dynamic compensation for polarization voltage hysteresis and electrolyte flow resistance effects, effectively solving the cross - scale modeling mismatch problem between electrochemical energy storage and mechanical energy storage systems; the collaborative mechanism of the multi - agent reinforcement learning framework and quantum - inspired optimization, which transforms the power distribution parameters and harmonic suppression objectives into quantum annealing constraint conditions, significantly enhancing the robustness of the converter modulation strategy against non - linear dynamic disturbances; the closed - loop energy - efficiency enhancement loop constructed by the federated reinforcement learning framework, which realizes the dynamic adaptation of the physical mechanism model and data - driven compensation through real - time feedback data - driven recalibration of the parameters of the neural differential equation by the edge computing node, and finally achieves the full - life - cycle energy - efficiency optimization of the energy storage system within the time range from seconds to hours, improving the dynamic response accuracy and energy conversion efficiency compared with traditional methods. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, for those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on the drawings.

[0044] Figure 1 It is a flowchart of an energy - saving optimization method for an electric energy storage device provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with the specific embodiments and corresponding drawings of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of them. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention. The following will describe in detail the technical solutions provided by each embodiment of the present invention in conjunction with the drawings. To better understand the objectives of the present invention, the present invention will be further described in detail below.

[0046] Please refer to Figure 1 , the present invention provides an energy - saving optimization method for an electric energy storage device, including:

[0047] Step S101, construct a multi-physics field coupling model of the energy storage device. By fusing the solid-liquid phase change kinetics equation of the lithium-ion battery, the fluid mechanics equation of the flow battery, and the rigid-flexible coupling equation of the flywheel rotor through neural differential equations, generate a dynamic meta-knowledge base containing cross-time-scale nonlinear interaction characteristics. The dynamic meta-knowledge base is a distributed database generated by fusing the multi-physics field coupling model through neural differential equations, storing cross-time-scale characteristics such as the solid-liquid phase change rate of the lithium-ion battery, the electrolyte flow velocity distribution of the flow battery, and the stress gradient of the flywheel rotor, and is used to provide spatio-temporal correlation constraint conditions for the physics-informed neural network;

[0048] The spatio-temporal attention mechanism includes a time attention layer and a space attention layer. The time attention layer analyzes the correlation between second-level charge-discharge data and hour-level aging data through a gated recurrent unit. The space attention layer captures the difference in the monomer voltage distribution within the battery stack through a convolutional kernel and fuses multi-modal features through a multi-head mechanism.

[0049] Step S102, based on the physical constraint conditions of the dynamic meta-knowledge base, use the physics-informed neural network to perform dynamic parameter compensation processing on the polarization voltage hysteresis effect and the electrolyte flow resistance coupling effect of the energy storage device, and output a residual feature vector containing the decoupled state-of-charge characteristics;

[0050] The Butler–Volmer electrochemical equation is a differential equation describing the kinetics of electrode reactions, and its expression is:

[0051]

[0052] i: current density, unit: A / m²;

[0053] : exchange current density, unit: A / m²;

[0054] α: transfer coefficient (dimensionless);

[0055] F: Faraday constant, value 96485 C / mol;

[0056] η: overpotential, unit: V;

[0057] R: gas constant, value 8.314 J / (mol·K);

[0058] T: temperature, unit: K.

[0059] Step S103, input the residual feature vector into the multi-agent reinforcement learning framework, generate the hierarchical control parameters of the supercapacitor terminal voltage and the magnetic bearing damping compensation amount of the flywheel energy storage, and form a cooperative control strategy;

[0060] Step S104: According to the power distribution threshold of the collaborative control strategy, transform the DC bus ripple suppression and harmonic distortion rate optimization objectives into the Hamiltonian constraint conditions of the quantum annealing model through the quantum-inspired optimization algorithm, and generate the converter modulation command. The quantum-inspired optimization algorithm specifically includes the following steps:

[0061] Map the DC bus ripple suppression objective to the local magnetic field strength of the qubit, transform the harmonic distortion rate optimization objective into the antiferromagnetic coupling relationship between adjacent qubits, and use the quantum annealing algorithm to solve the ground state of the Hamiltonian to generate the converter modulation command.

[0062] Step S105: In the federated reinforcement learning framework, distill the cloud global optimization strategy to the edge controller through the imitation learning generative adversarial network, and at the same time inject the real-time working condition data fed back by the edge controller into the multi-physics field coupling model for parameter recalibration, forming a closed-loop energy efficiency enhancement loop that crosses physical models, data-driven, and quantum optimization.

[0063] When constructing the multi-physics field coupling model of the energy storage device, use the neural differential equation to couple and solve the kinetic equation of the solid-liquid phase change process of the lithium-ion battery and the Navier-Stokes fluid control equation of the flow battery. At the same time, introduce the rigid-flexible coupling vibration equation of the flywheel rotor, and extract the characteristics of the temperature field, flow velocity field, and stress field data at multiple time scales through the variational autoencoder. This process generates a dynamic meta-knowledge base containing the correlation characteristics of the second-level charge and discharge response and the hourly cycle decay, and the meta-knowledge base is stored in the distributed graph database for providing cross-domain constraint conditions for subsequent physical parameter compensation.

[0064] Based on the physical constraint conditions of the dynamic meta-knowledge base, design a physics-informed neural network on the lithium-ion battery side, embed the Butler–Volmer electrode reaction equation into the long short-term memory (LSTM) network in a hard constraint manner, and construct a state-of-charge estimation model with electrochemical mechanism; on the flow battery side, use a graph convolutional network to model the electrolyte flow channel topology, combine the federated learning framework to aggregate the data of multiple pumping station pressure sensors, and generate the electrolyte flow velocity compensation matrix. The outputs of the two compensation mechanisms are fused through the feature splicing layer into a residual vector containing the decoupled state of charge and flow resistance characteristics, and its dimension matches the input layer structure of the multi-agent reinforcement learning framework.

[0065] After the residual feature vector is input into the multi-agent reinforcement learning framework, the supercapacitor control branch uses a hierarchical attention mechanism to analyze the charge relaxation process, and dynamically adjusts the sliding mode control surface parameters through the double-delay deep deterministic policy gradient algorithm; the flywheel energy storage branch deploys a rotor deformation observer based on neural radiance fields, and uses the proximal policy optimization algorithm to generate the active damping compensation amount of the magnetic bearing. The outputs of the two branches are fused by the policy fusion layer to generate the power distribution command and the damping control signal, forming a cooperative control strategy that meets the requirements of multi-time scale response.

[0066] The power distribution threshold of the cooperative control strategy is mapped to the Ising model Hamiltonian through the quantum approximate optimization algorithm, and the objectives of suppressing the DC bus ripple and optimizing the harmonic distortion rate are encoded as the ground state search problem of the spin glass system. During the quantum annealing process, the transverse magnetic field strength is dynamically adjusted by the adaptive tuning algorithm, and combined with the harmonic spectrum distribution characteristics predicted by the deep residual network, a carrier phase-shifted modulation instruction set that meets the multi-objective Pareto optimality is generated.

[0067] In the federated reinforcement learning framework, the multi-agent proximal policy optimization algorithm deployed in the cloud fuses the real-time working condition data uploaded by the edge nodes through the differential privacy mechanism to generate a global optimization strategy; the edge side uses the generative adversarial network architecture for policy distillation to simplify the complex policy into a lightweight control instruction. The working condition data collected after the edge controller executes the instruction is injected into the multi-physical field coupling model through the backpropagation channel, triggering the online calibration of the boundary conditions in the neural differential equation, and forming a closed-loop enhancement loop covering physical mechanism modeling, data-driven compensation, and quantum optimization modulation. This loop realizes the continuous optimization of the energy efficiency during the whole life cycle of the energy storage system through the parameter iterative update mechanism of the dynamic meta-knowledge base.

[0068] Specifically, for the energy-saving optimization method of the electric energy storage device described in the present invention, the construction of the multi-physical field coupling model of the energy storage device includes:

[0069] Using neural differential equations to fuse the solid-liquid phase change reaction kinetics equation of lithium-ion batteries, the hydrodynamics equation of flow batteries, and the rigid-flexible coupling equation of flywheel rotors;

[0070] Extracting the spatio-temporal correlation features of multi-source heterogeneous data of the energy storage device through the Transformer architecture with spatio-temporal attention mechanism, generating a meta-knowledge base containing cross-scale non-linear interaction relationships, and the meta-knowledge base is used to provide physical constraint conditions for the physical information neural network.

[0071] When constructing the multi-physics coupling model of the energy storage device, first, the solid-liquid phase change reaction kinetic equation of the lithium-ion battery is expressed in the form of a partial differential equation, and the continuous-time neural network architecture in the neural differential equation is used for discretization, and the phase change interface evolution process is solved by the adaptive step-size algorithm. At the same time, the spectral method is used for spatial discretization of the fluid mechanics equation of the flow battery, and the Reynolds-averaged Navier-Stokes model is combined to capture the turbulence effect. The rigid-flexible coupling equation of the flywheel rotor is constructed by the Hamilton variational principle, and the symplectic algorithm is used to maintain the energy conservation characteristics of the numerical calculation. The above three types of equations are coupled through a multi-task learning framework with a shared hidden layer, and the attention mechanism is used to dynamically allocate the weight coefficients of different physical fields at the coupling boundary to form a unified differential operator that can represent the cross-time-scale interaction.

[0072] During the solution process of the neural differential equation, a variational autoencoder is used to compress the features of the multi-physics output, and the key modal features of the solid-liquid phase change rate, electrolyte flow rate, and rotor stress distribution are extracted. The feature decoupling ability is enhanced through adversarial generation training, so that the latent variables output by the encoder respectively correspond to the independent action components of chemical kinetics, fluid mechanics, and mechanical dynamics. The finally generated coupling model parameters are stored in the distributed parameter server to form a dynamic meta-knowledge base containing the multi-physics coupling relationship.

[0073] Based on the above coupling model, the operating data of the energy storage device is processed through the Transformer architecture. A spatio-temporal attention mechanism is constructed, in which the time attention layer analyzes the correlation pattern between the second-level charge-discharge data and the hourly cycle aging data, and the spatial attention layer captures the monomer distribution difference in the battery stack and the flow channel pressure drop gradient of the flow battery. The multi-head attention mechanism is used to process the temperature sensor data, voltage and current time series data, and vibration spectrum data in parallel, and a tensor representation with spatio-temporal correlation is generated through cross-modal feature crossing. The tensor is input into the feed-forward neural network for feature recombination to generate a meta-knowledge graph containing cross-scale non-linear interaction relationships.

[0074] The meta-knowledge graph is lightweight processed by the knowledge distillation technology, and node embedding vectors are constructed through the graph neural network, where each node corresponds to a specific physical constraint condition. The edge weights in the graph are dynamically matched with the residual connection layer parameters of the physical information neural network, so that the flow channel optimization constraints of the flow battery are automatically mapped to the update rule of the adjacency matrix of the graph convolutional network, and the phase change constraints of the lithium-ion battery are associated with the forgetting gate control mechanism of the long short-term memory network. Through the online incremental learning mechanism, the newly discovered physical constraint relationships during the operation of the energy storage device are continuously supplemented to the meta-knowledge base in real time to form a continuously evolving cross-domain constraint supply system.

[0075] Specifically, for the energy-saving optimization method of the electrical energy storage device described in the present invention, the dynamic parameter compensation includes:

[0076] On the lithium-ion battery side of the energy storage device, the Butler-Volmer electrochemical equation is embedded into the long short-term memory network to decouple the cross-sensitivity between the state of charge and the state of health in real time;

[0077] On the flow battery side, a graph neural network is constructed to predict the pressure drop distribution in the flow channel. Combining federated learning to aggregate the operating data of multi-node pump units, the feedforward compensation coefficient matrix of the electrolyte flow rate is dynamically optimized, and the compensation residual is used as the state observation input of the multi-agent reinforcement learning. The specific federated learning framework includes: the edge nodes train the graph neural network model through the local dataset to generate model gradients; the central server aggregates the gradients using the dynamic weighted average algorithm to update the global model parameters; Gaussian noise is added to the gradients through the differential privacy mechanism.

[0078] In the parameter compensation process of the lithium-ion battery, the activation overpotential term and the concentration overpotential term of the Butler-Volmer electrochemical equation are respectively mapped to the input gate and the forget gate of the long short-term memory network, and the temporal modeling of the electro-chemical reaction kinetic characteristics is realized through the gating mechanism. The residual connection structure is used to separate the phase change characteristics of the electrode material and the diffusion characteristics of the electrolyte, and a two-channel feature extraction network is constructed. The main channel learns the dynamic change mode of the state of charge, and the auxiliary channel captures the degradation trajectory of the state of health. The outputs of the two channels are processed by orthogonalization constraints, and the cross-sensitivity between state variables is eliminated through covariance matrix decomposition to generate the decoupled state of charge estimation value and the state of health degradation rate parameter.

[0079] For the prediction of the pressure drop in the flow channel of the flow battery, a flow channel topology structure model is constructed based on the graph neural network. The electrolyte flow channel is decomposed into a set of nodes and edges. The node attributes include the flow rate, pressure, and temperature sensor data, and the edge weight represents the cross-sectional change rate and the bending angle of the flow channel. The message passing mechanism is used to aggregate the features of adjacent nodes, and the heat map of the pressure drop distribution in the flow channel is generated through multi-layer graph convolution operations. Combining the federated learning framework, the global model is deployed on the central server, receiving the pump speed, valve opening, and pressure difference data uploaded by each edge node, and using the dynamic weighted average algorithm to update the weight parameters of the graph neural network to generate a feedforward compensation coefficient matrix adapted to the multi-node working conditions.

[0080] The feedforward compensation coefficient matrix is dynamically optimized through an online incremental learning mechanism. The compensation residual is defined as the integral of the temporal error between the actual flow rate and the predicted value. The sliding window mechanism is used to extract the frequency domain features of the residual signal. The residual feature vector and the decoupling parameters of the lithium-ion battery are processed for spatio-temporal alignment to construct a multi-dimensional state observation tensor, whose dimension matches the input layer structure of the multi-agent reinforcement learning framework. After being normalized, this tensor is input into the policy network to provide real-time state feedback for the power distribution of the supercapacitor and the flywheel damping control.

[0081] During the multimodal fusion process to compensate for residual errors, a feature attention mechanism is designed to dynamically assign weight coefficients to the parameters of lithium-ion batteries and flow batteries. A differentiable decision tree is used to analyze the correlation between residual features and system energy efficiency, generating feature importance scores. Based on these scores, the feature combination of state observation vectors in the multi-agent reinforcement learning framework is dynamically adjusted. The fused feature vectors are then passed through a temporal convolutional network to extract long-term and short-term dependencies, providing a foundation for cross-timescale state representation for collaborative control strategy generation.

[0082] Specifically, the energy-saving optimization method for an electric energy storage device according to the present invention inputs the residual feature vector into a multi-agent reinforcement learning framework to generate hierarchical control parameters for the supercapacitor terminal voltage and the damping compensation amount of the flywheel energy storage magnetic bearing, thereby forming a collaborative control strategy including:

[0083] A hierarchical attention mechanism is used to control the supercapacitor terminal voltage, and a sliding mode parameter dynamic adjustment strategy is used to match the charge relaxation time constant.

[0084] A rotor deformation observer based on neural radiation field is deployed on the flywheel energy storage side, and combined with the proximal strategy optimization algorithm to generate the magnetic bearing damping compensation, so that the quantum heuristic optimization algorithm receives the torque fluctuation suppression threshold as the modulation target constraint.

[0085] In supercapacitor terminal voltage control, a hierarchical attention mechanism is constructed, which includes both spatial and temporal feature processing. The spatial attention layer analyzes the differences in cell voltage distribution within the battery stack and generates a region-sensitive weight matrix using convolution kernels. The temporal attention layer extracts the relaxation time characteristics of the charge and discharge curves, using gated recurrent units to capture millisecond-level transient responses and second-level steady-state transition modes. After feature fusion of the two attention layers' outputs, dynamic sliding mode surface parameter adjustment instructions are generated. These instructions use adaptive laws to match the time-varying characteristics of the supercapacitor's double-layer charge relaxation process in real time, dynamically aligning the sliding mode control bandwidth with the charge migration rate.

[0086] Based on the dynamic sliding mode surface parameters, a sliding mode observer is designed to estimate the residual capacity deviation during charge relaxation in real time. The sliding mode reaching law is derived through Lyapunov stability analysis. The observer output is coupled with the modulation frequency parameter of a quantum-inspired optimization algorithm to establish a quantitative relationship between terminal voltage control accuracy and converter switching losses. The observer residual signal is fed back into a hierarchical attention mechanism to form a parameter self-correction loop.

[0087] On the flywheel energy storage side, a rotor deformation observer based on neural radiance fields is deployed. The deformation data of the rotor surface is collected by a multi-view laser displacement sensor, and a three-dimensional point cloud spatial distribution model is constructed. The implicit neural representation method is used to encode the point cloud data into a continuous radiance field function, and the dynamic deformation surface of the rotor is reconstructed by volume rendering technology. The deformation gradient tensor is input into the value function network of the proximal policy optimization algorithm to calculate the torque fluctuation components in the directions of each degree of freedom.

[0088] The proximal policy optimization algorithm adopts a trust region constraint mechanism to generate the multi-phase winding current compensation amount of the magnetic bearing according to the torque fluctuation spectrum distribution characteristics. During the compensation amount generation process, the smoothness constraint of the compensation instruction is maintained by the policy gradient projection algorithm to avoid introducing high-frequency oscillation components into the second harmonic. The generated damping compensation amount threshold is used as a boundary condition to input the quantum annealing optimization process, which restricts the search space of the carrier phase shift angle of the converter, so that the harmonic suppression frequency band of the modulation strategy forms a dynamic avoidance area with the resonance frequency band of the flywheel rotor.

[0089] During the execution of the coordinated control strategy, the power distribution coefficient controlled by the output voltage of the supercapacitor is coupled with the flywheel damping compensation amount threshold through the feature cross-attention mechanism. The cross-attention layer analyzes the cooperative relationship between the two in the millisecond-level transient response and second-level steady-state maintenance stages, and generates a joint optimization target weight matrix. The matrix is input into the Hamiltonian construction module of the quantum-inspired optimization algorithm to drive the multi-objective Pareto front search process to converge to the optimal solution of the system-level energy efficiency.

[0090] Specifically, for the energy-saving optimization method of the electric energy storage device described in the present invention, the multi-objective optimization includes:

[0091] Converting the DC bus ripple suppression, harmonic distortion rate optimization, and switching loss minimization into the Hamiltonian of the quantum annealing model;

[0092] Predicting the harmonic spectrum distribution characteristics through a deep residual network, and feeding the distribution characteristics back to the multi-physics field coupling model to update the boundary conditions of the neural differential equation.

[0093] During the construction of the quantum annealing model, the DC bus ripple suppression target is encoded as the local field strength parameter of the spin system, and its strength is inversely proportional to the effective value of the ripple voltage; the harmonic distortion rate optimization target is mapped to the interaction coefficient between adjacent spins, and the corresponding relationship between the harmonic order and the coupling strength is constructed through the topological sorting algorithm; the switching loss minimization target is converted into the system ground state energy constraint condition, and the Lagrange multiplier method is used to balance the loss weight and the ripple suppression requirement. The three types of optimization targets are normalized to form a composite Hamiltonian, which is input into the quantum annealing solver for ground state search.

[0094] The deep residual network (DRN) employs a multi-scale convolutional kernel structure. The input layer receives converter output current waveform data and extracts joint time-frequency domain features by stacking residual blocks. A skip connection structure fuses shallow, detailed features with deep semantic features. The output layer generates a harmonic spectrum distribution vector, containing the amplitude and phase information of each harmonic and interharmonic clustering features. Adversarial sample enhancement technology is used during network training to improve robustness to background harmonic interference from the power grid.

[0095] The harmonic spectrum distribution vector is extracted through a feature selection module to extract the energy contribution parameters of key frequency bands. This is then weighted using an attention mechanism to generate a boundary condition update vector. This vector is then fed into the neural differential equation solver of the multiphysics coupling model to adjust the viscosity boundary values in the flow battery fluid dynamics equations and correct the interfacial energy parameters in the lithium-ion battery phase transition equations. The updated boundary conditions trigger a recalculation of the coupled model, generating a new set of dynamic meta-knowledge base parameters.

[0096] The optimal spin configuration output by quantum annealing is decoded into a sequence of carrier phase-shift angles. Combined with the distribution of harmonic-sensitive frequency bands predicted by a deep residual network, the dead time and switching frequency of the pulse-width modulation wave are dynamically adjusted. Current waveform data collected during the modulation strategy execution forms a closed-loop feedback loop. The residual network convolution kernel weights are updated through an online learning mechanism, enabling the harmonic prediction model to continuously adapt to characteristic drift caused by converter aging.

[0097] The interaction between multi-objective optimization results and the collaborative control strategy is achieved through a policy fusion layer, which spatially and temporally aligns the modulation instructions generated by quantum annealing with the power allocation parameters of multi-agent reinforcement learning. During this alignment, a dynamic time warping algorithm is used to compensate for control delays, generating a joint optimization instruction set to drive the energy storage device. Simultaneously, execution performance data is fed into the federated learning framework for global policy updates.

[0098] Specifically, the energy-saving optimization method for an electric energy storage device according to the present invention, wherein the closed-loop energy efficiency enhancement circuit across physical models, data drive, and quantum optimization is formed, comprises:

[0099] Based on the output of the multi-agent reinforcement learning framework, a multi-agent proximal strategy optimization algorithm is deployed in the cloud, and the multi-node pump operation data aggregated by the federated learning is integrated through the differential privacy protection mechanism;

[0100] On the edge side, an imitation learning generative adversarial network is used to convert cloud strategies into lightweight control instructions, and the instruction execution results are reversely injected into the generated dynamic meta-knowledge base to trigger the recalibration of the boundary condition parameters of the neural differential equation.

[0101] When deploying the multi-agent proximal policy optimization algorithm in the cloud, a hierarchical policy network structure is constructed. The global policy network receives the multi-node pump operation data aggregated from the federated learning framework, including the time-series information of pressure, flow rate, and temperature. The differential privacy protection mechanism is used to add noise to the data, and the privacy budget parameter is dynamically adjusted to balance the data utility and the privacy protection intensity. The output layer of the policy network generates the global optimization policy containing the power distribution coefficient and the frequency adjustment instruction, which is encrypted and transmitted to the edge node through the secure multi-party computation protocol.

[0102] The generative adversarial network deployed on the edge side includes a policy generator and a discriminator module. After receiving the encrypted policy from the cloud, the generator uses the knowledge distillation technology to extract key features and reconstructs them into a control instruction set suitable for edge computing resources through a lightweight convolutional neural network. The discriminator module compares the action value distributions of the reconstructed instruction and the original policy, and generates an adversarial loss function to guide the optimization of the generator. The generated lightweight instructions include the charge and discharge thresholds of the supercapacitor and the flywheel speed control parameters, which are converted into PWM modulation signals by the edge controller to drive the power module.

[0103] The instruction execution result data is collected by the edge computing node, including the bus voltage fluctuation spectrum, the energy storage unit efficiency parameter, and the device temperature gradient. After feature extraction, the collected data is used to construct a feedback vector, which is transmitted to the dynamic meta-knowledge base storage layer through the secure tunnel protocol. The feedback vector triggers the version control mechanism of the meta-knowledge base, and filters out the historical parameter set with the highest correlation with the current working condition for comparison and analysis.

[0104] During the re-calibration of the boundary conditions of the neural differential equation, the Bayesian optimization algorithm is used to calculate the deviation between the feedback vector and the current boundary conditions, and the viscosity coefficient tensor of the Navier-Stokes equation of the flow battery and the flywheel rotor stress relaxation parameters are updated by the gradient descent method. The calibrated differential equation parameters are injected into the online solver of the multi-physics field coupling model to generate updated cross-time scale interaction features, forming a real-time state mirror of the digital twin.

[0105] During the closed-loop optimization process, the updated dynamic meta-knowledge base parameters are synchronized to the value function estimator of the cloud policy network to adjust the balance coefficient of policy exploration and exploitation. At the same time, the calibrated boundary conditions are injected into the Hamiltonian construction module of the quantum optimization algorithm as physical constraints to constrain the search space of the converter modulation strategy, forming a cross-layer optimization architecture for cloud-edge collaboration. This architecture improves the two-way mapping accuracy between the digital twin and the physical system through the continuous evolution of the dynamic meta-knowledge base.

[0106] Specifically, for the energy-saving optimization method of the electric energy storage device described in the present invention, the federated reinforcement learning framework includes:

[0107] Based on the physical constraints of the dynamic meta-knowledge base, a hierarchical reward mechanism is designed to map the dynamic efficiency of the energy storage system to the depth of the potential energy well, and a long-term energy efficiency optimization goal is constructed.

[0108] A dynamic energy efficiency entropy index is generated by a multi-entropy maximization controller, and this index is used to constrain the charge relaxation threshold in the power distribution range of the supercapacitor in the generated cooperative control strategy.

[0109] When constructing the hierarchical reward mechanism, based on the phase change kinetic parameters of lithium-ion batteries and the flow resistance characteristics of flow batteries in the dynamic meta-knowledge base, a potential energy well depth calculation module is designed. The instantaneous efficiency index of the energy storage system is mapped to the gradient change rate of the potential energy function surface, where the local extreme point of the potential energy well corresponds to the optimal working range. An energy conservation equation for the long-term energy efficiency optimization goal is generated by the Hamiltonian Monte Carlo sampling method, where the kinetic energy term characterizes the power regulation rate and the potential energy term reflects the cumulative loss of the state of charge deviating from the optimal range. The potential energy well parameters are dynamically updated through an online learning mechanism and kept synchronized with the global model parameters in the federated learning framework.

[0110] In the construction process of the multi-entropy maximization controller, a differentiable logic programming is used to define the dynamic energy efficiency entropy index. The input layer receives the double-layer voltage distribution data of the supercapacitor and the stress tensor of the flywheel rotor, and extracts the joint time-frequency domain features of the charge relaxation process through a feature crossing network. A causal graph model is constructed to analyze the intervention effect of the power distribution action and the charge relaxation time, and a causal intervention reward term is generated to eliminate the spurious correlation between the environmental state and the action selection. The entropy value calculation module combines the Shannon entropy and the Rayleigh entropy to construct a hybrid metric index to constrain the charge migration path of the power distribution strategy in the transient response stage.

[0111] The dynamic energy efficiency entropy index acts on the cooperative control strategy through a policy distillation mechanism, and an entropy constraint module is embedded in the supercapacitor terminal voltage control loop. This module monitors the charge relaxation time constant in the power distribution range in real time. When the oversaturation trend of the double layer is detected, a sliding mode parameter adaptive adjustment mechanism is triggered. During the adjustment process, the modulation frequency parameters output by the quantum optimization algorithm are combined to dynamically reconstruct the boundary conditions of the sliding mode control surface, so that the charge relaxation process is always within the safe threshold range of the entropy value constraint.

[0112] The cooperation between the hierarchical reward mechanism and the multi-entropy controller is realized through the federated learning architecture. The cloud policy network receives the potential energy well parameters and entropy value distribution data from multiple nodes, and uses an attention mechanism to generate a weight coefficient matrix. After receiving the global weight coefficients, the edge-side controller converts the entropy constraint conditions into the feasible domain boundary of the local control instructions through a policy projection algorithm. The characteristic data of the charge relaxation process collected during the execution of the instructions are uploaded to the dynamic meta-knowledge base through the differential privacy mechanism, triggering the online calibration of the curvature parameters of the potential energy function surface.

[0113] During the calibration of the potential energy function, an adversarial generative network is used to synthesize charge migration path data under critical operating conditions, and the representation ability of the potential energy well for abnormal operating conditions is enhanced through a contrastive learning mechanism. The calibrated potential energy parameters are fed back to the multi-entropy controller to adjust the intervention effect intensity coefficient in the causal graph model, forming a dynamic collaborative optimization loop of the reward mechanism and the control strategy. This loop continuously iterates to improve the adaptability of the supercapacitor power distribution strategy to the charge relaxation process, achieving the optimal energy efficiency control of the energy storage system's transient response.

[0114] Specifically, the energy-saving optimization method of the electric energy storage device described in the present invention further includes:

[0115] Embedding an adversarial curriculum generator within the federated learning framework to synthesize training scenarios based on the critical operating condition data stream;

[0116] The federated learning framework performs differential privacy processing on the multi-node pump operation data to generate differentially private protected data;

[0117] Evaluating the distribution difference degree between the synthesized scenarios and the differentially private protected data through a graph attention network, and dynamically adjusting the training curriculum difficulty coefficient of the multi-agent reinforcement learning framework.

[0118] When constructing an adversarial curriculum generator within the federated learning framework, a Wasserstein generative adversarial network architecture is used to design the scenario synthesis module. The generator receives the feature vectors from the critical operating condition data stream, including the time series of grid frequency mutation events and the spectral characteristics of load step responses, and generates synthetic scenario data with high information entropy through a deep convolutional network. The discriminator adopts a graph attention network structure, where the node attributes include the voltage volatility, temperature gradient, and vibration spectrum parameters of the real operating conditions, and the edge weights represent the Markov transition probability between the operating condition characteristics. The Wasserstein distance between the synthetic scenarios and the differentially private protected data is calculated through a multi-head attention mechanism.

[0119] The graph attention network sets a hierarchical attention mechanism. The first-layer attention heads analyze the statistical distribution differences of the node attributes, and the second-layer attention heads capture the topological structure differences of the edge weights. The difference degree calculation module weights and fuses the outputs of each attention head to generate a scalar index representing the fidelity of the synthetic scenarios. This index is input into the curriculum difficulty regulator, and the mixing ratio of the synthetic scenarios and the real data in the training curriculum is dynamically adjusted through a fuzzy logic controller, where the difficulty coefficient is negatively correlated with the exploration rate parameter of the agent policy network.

[0120] When the dynamically adjusted training course data is input into the multi-agent reinforcement learning framework, a course progressive training strategy is adopted. In the initial stage, strongly perturbed scenario data is injected with a high difficulty coefficient to activate the exploration mechanism of the policy network. As the training process progresses, the proportion of synthetic scenarios is gradually reduced and the weight of real data is increased. The value function estimator of the policy network receives the difficulty coefficient as prior knowledge and adjusts the uncertainty range of value estimation through a Bayesian neural network to balance the optimization direction of exploration and exploitation.

[0121] The collaboration between the adversarial course generator and the federated learning framework is achieved through a dual feedback mechanism. On the one hand, the performance data of the multi-agent policy network in the synthetic scenario is fed back to the reward signal generation module of the generator, driving the generator to optimize the criticality features of the scenario. On the other hand, the distribution difference data evaluated by the graph attention network is injected into the model update rule of the federated aggregation server to dynamically adjust the model contribution weights of each edge node. This mechanism enables the evolution direction of the synthetic scenario to be dynamically aligned with the real working condition distribution, while promoting the improvement of the robustness of the global policy.

[0122] The dynamic adjustment process of the course difficulty coefficient forms a closed-loop linkage with the parameter calibration of the digital twin. When it is detected that the real-time working condition data fed back by the edge controller exceeds the current course coverage range, a step adjustment instruction for the course difficulty coefficient is triggered. This instruction is synchronously sent to the re-initialization module of the cloud policy network, and the adaptability of the policy network to the new working condition mode is enhanced through a partial parameter reset mechanism. At the same time, the boundary condition constraint set of the neural differential equation is updated to maintain the mirror accuracy between the digital twin and the physical system.

[0123] Specifically, the energy-saving optimization method of the electric energy storage device described in the present invention further includes:

[0124] Construct an unsteady Pareto front search engine, and dynamically allocate multi-objective optimization weights according to the dynamic energy efficiency entropy index and the harmonic spectrum distribution characteristics;

[0125] Input the weight sequence into the quantum-inspired optimization algorithm, and update the Hamiltonian constraint conditions of the quantum annealing model to achieve the real-time alignment of the converter modulation strategy and the dynamic energy efficiency optimization goal.

[0126] When constructing the unsteady Pareto front search engine, an improved non-dominated sorting genetic algorithm framework is adopted. The input layer receives the time series data of the dynamic energy efficiency entropy index and the frequency domain feature vector of the harmonic spectrum distribution. The long short-term memory network is used to predict the evolution trend of the multi-objective weight vector, and the prediction result is encoded as the crossover probability parameter of the gene sequence. Set the evolutionary strategy guided by the hypervolume index, and use the Monte Carlo tree search algorithm to select the mutation operator in the strategy space that can simultaneously expand the hypervolume of the objective function and reduce the conflict degree, and generate the dynamic topological structure of the non-dominated solution set.

[0127] During the dynamic weight allocation process, a feature attention mechanism is constructed to analyze the correlation intensity between the energy efficiency entropy index and the harmonic features, and the contribution degree scores of each objective function are calculated through a differentiable decision tree. The scoring results are input into a fuzzy logic controller to generate an adaptive adjustment coefficient matrix of the weight vector, where the weight values in the harmonic sensitive frequency band form a non-linear mapping relationship with the transverse magnetic field strength of the quantum annealing model. The weight sequence is synchronized with the converter switching period through a spatio-temporal alignment module to generate an optimized instruction stream with timestamp matching.

[0128] When the weight sequence is input into the quantum-inspired optimization algorithm, the spin coding technology is used to map the weight values to the coupling strength parameters of the quantum bits. The DC bus ripple suppression target is encoded as the local magnetic field bias, the harmonic distortion rate optimization target is transformed into the antiferromagnetic coupling relationship between adjacent quantum bits, and the switching loss constraint is embodied as the system ground state energy threshold. The updated Hamiltonian constraint conditions are solved through the heat bath algorithm of the quantum annealer, and an optimized sequence of carrier phase-shifting angles that satisfies multi-objective trade-off is output.

[0129] The real-time alignment process of the converter modulation strategy adopts a sliding window mechanism to couple the angle sequence output by the quantum optimization and the power allocation instructions generated by the multi-agent reinforcement learning framework in space and time. The transmission delay of the control instructions is compensated through the dynamic time warping algorithm to generate a phase-synchronized pulse width modulation waveform. The current harmonic data collected during the execution of the modulation strategy is fed back to the objective function update module of the Pareto front search engine after fast Fourier transform to form a closed-loop optimization link.

[0130] In the closed-loop feedback mechanism, the real-time change data of the harmonic spectrum characteristics triggers the topological structure reorganization of the Pareto front, and the priority rules of non-dominated sorting are dynamically adjusted through the gradient boosting tree algorithm. The reorganized front solution set is distributed to each edge node through the federated learning framework to update the Hamiltonian constraint parameters of the local quantum optimization model. At the same time, the optimization deviation is injected into the dynamic meta-knowledge base to trigger the recalibration operation of the neural differential equation, forming a collaborative enhancement effect across optimization levels.

[0131] Specifically, for the energy-saving optimization method of the electric energy storage device described in the present invention, the closed-loop energy efficiency enhancement loop includes:

[0132] Inject the synthesized critical condition data stream reversely into the dynamic meta-knowledge base to trigger the reparameterization of the hydrodynamic equation of the flow battery in the neural differential equation;

[0133] Dynamically constrain the frequency response interval of the flywheel energy storage magnetic bearing damping compensation amount in the generated cooperative control strategy through the hierarchical reward mechanism and the dynamic energy efficiency entropy index;

[0134] The updated dynamic meta-knowledge base parameters are synchronously fed back to the physical information neural network and quantum optimization algorithm to form a closed-loop enhancement circuit for cross-model parameter compensation, control strategy generation and modulation optimization.

[0135] During the reverse injection of critical operating condition data streams, a generative adversarial network is employed to enhance the feature diversity of the synthesized data, encoding the time-frequency domain features of scenarios such as grid frequency drops and sudden load changes into high-dimensional tensors. The data stream is transmitted via a distributed message queue to the incremental learning module of the dynamic meta-knowledge base, triggering the reparameterization of the flow battery's Navier-Stokes equations. This reparameterization utilizes a Bayesian optimization algorithm to calculate corrections to the fluid viscosity and turbulence model. An online learning mechanism is then used to update the parameters of the vortex dissipation term in the differential equations, generating a fluid dynamics characteristic matrix adapted to extreme operating conditions.

[0136] The calculation module for the dynamic energy efficiency entropy index within the hierarchical reward mechanism receives flywheel rotor vibration spectrum data and uses wavelet packet decomposition to extract the energy distribution characteristics of the resonant frequency band. A dynamic weight allocation algorithm decomposes the entropy index into a fundamental frequency constraint term and a harmonic suppression term. The fundamental frequency constraint term maps to the upper frequency response limit of the magnetic bearing current compensation, while the harmonic suppression term corresponds to the phase delay threshold of the damping compensation. A feature attention mechanism is constructed to analyze the correlation pattern between the compensation threshold and the rotor stress tensor, generating dynamic boundary conditions for the frequency response range.

[0137] When frequency response boundary conditions are input into the collaborative control strategy generation module, a hybrid architecture combining sliding mode control and adaptive fuzzy logic is employed. The sliding surface parameters are dynamically adjusted based on the real-time frequency response error. The fuzzy rule base receives the modulation frequency parameters output by the quantum optimization algorithm and generates compensation instructions for smooth transitions. Flywheel speed fluctuation data collected during the execution of the compensation instructions is extracted using a fast Fourier transform to extract frequency domain features. This data is then fed back into the hierarchical reward mechanism to update the entropy index calculation weights.

[0138] The updated dynamic meta-knowledge base parameters are synchronized with the physical information neural network and quantum optimization algorithm via a version control protocol. The physical information neural network employs a transfer learning strategy to align the reparameterized fluid dynamics characteristic matrix with the lithium-ion battery phase transition model across domains, updating the electrochemical-fluid coupling compensation parameters. After receiving the synchronized parameters, the quantum optimization algorithm reconstructs the spin coupling coefficients in the Hamiltonian constraints and adjusts the ground state search path using an annealing temperature scheduling algorithm.

[0139] During the operation of the closed-loop enhancement circuit, the edge computing node collects converter modulation waveforms and energy storage unit efficiency data in real time, extracts features, and generates feedback vectors. These feedback vectors are uploaded to the cloud-based optimization engine via the federated learning framework's aggregation channel, triggering iterations of the dynamic meta-knowledge base and updates to the policy network parameters. The updated control strategy is lightweight and compressed before being distributed to the edge controller, forming a full-link optimization system spanning physical modeling, collaborative control, and quantum modulation, achieving continuous improvements in the energy efficiency of the energy storage system and its equipment lifespan.

[0140] Specific embodiments of the present invention relate to multi-level optimization control of electric energy storage devices. In response to the energy efficiency improvement needs of lithium-ion batteries, liquid flow batteries, supercapacitors and flywheel energy storage systems under complex working conditions, dynamic optimization is achieved through the deep integration of physical models and data-driven technologies. In the multi-physics field coupling modeling stage of the energy storage device, the neural differential equation is used to implicitly discretize and solve the solid-liquid phase change kinetic equation of the lithium-ion battery, and the adaptive step size algorithm is combined to capture the evolution characteristics of the interface of the electrode active material; the Navier-Stokes fluid equation of the liquid flow battery is spatially discretized using the spectral element method, and the turbulent vortex dissipation effect is predicted by the Reynolds stress model; the flywheel rotor rigid-flexible coupling equation is constructed based on the Hamiltonian variational principle, and the symplectic algorithm is used to maintain the mechanical energy conservation characteristics. The three types of equations are coupled through a multi-task learning framework with a shared hidden layer, and the attention mechanism is used to dynamically allocate the coupling boundary weights to generate a dynamic meta-knowledge base containing the correlation characteristics of second-level transient response and hour-level cyclic attenuation.

[0141] During dynamic parameter compensation, the activation overpotential term of the Butler–Folmer equation is embedded in the input gated unit of the long-short-term memory network on the lithium-ion battery side. A residual connection structure is used to separate electrode phase change characteristics from electrolyte diffusion effects, and covariance matrix decomposition is used to eliminate cross-sensitivity between state of charge and state of health. On the flow battery side, a flow channel topology model is constructed based on a graph neural network. Node attributes integrate pressure sensor data and infrared temperature field distributions. A federated learning framework aggregates operating data from multiple pumping stations to generate an electrolyte flow rate compensation matrix. The compensation residuals are then subjected to a sliding window mechanism to extract frequency domain features, and a multi-dimensional state observation tensor is constructed as input into a multi-agent reinforcement learning framework.

[0142] During the collaborative control strategy generation phase, the supercapacitor voltage control employs a hierarchical attention mechanism to analyze the double-layer relaxation process. The spatial attention layer uses convolution kernels to identify differences in voltage distribution within the battery stack, while the temporal attention layer employs gated recurrent units to capture the transient response patterns of charge migration. The flywheel energy storage side reconstructs the rotor deformation surface based on the neural radiation field and generates current compensation for the multiphase windings of the magnetic bearing using a proximal strategy optimization algorithm. The compensation threshold serves as a torque fluctuation constraint for the quantum optimization algorithm. The strategy fusion layer spatially and temporally aligns the power allocation parameters with the damping compensation to generate a joint optimization instruction set.

[0143] In the quantum-inspired optimization stage, the DC bus ripple suppression target is encoded as the local field strength parameter of the spin system, and the harmonic distortion rate optimization is mapped to the antiferromagnetic coupling relationship between qubits. The time-frequency joint features of the current waveform are extracted through a deep residual network to predict the distribution of harmonic-sensitive frequency bands. Based on the updated Hamiltonian constraint conditions, the quantum annealer outputs a carrier phase-shift angle sequence, and combines with the dynamic time warping algorithm to compensate for the control delay to generate a phase-synchronized pulse-width modulation waveform.

[0144] In the federated reinforcement learning framework, the cloud policy network fuses the pump pressure data of multiple nodes through the differential privacy mechanism to generate a global optimization strategy; on the edge side, a generative adversarial network is used for policy distillation to simplify the complex control logic into lightweight PWM instructions. The instruction execution data triggers the recalibration of the boundary conditions of the neural differential equation through the backpropagation channel to update the viscosity coefficient and interfacial energy parameters of the dynamic meta-knowledge base. The closed-loop optimization loop synchronizes the cloud and edge parameters through the version control protocol to realize the dynamic mirroring of the digital twin and the physical system, and finally achieves the improvement of the energy efficiency of the entire life cycle of the energy storage system.

[0145] The technical solution of the present invention solves the problems of model mismatch and control deviation of the energy storage device through three-level technical means: constructing a dynamic coupling model across time scales, establishing a data-driven parameter compensation mechanism, and realizing closed-loop optimization control. First, a neural differential equation is used to fuse the phase change dynamics of lithium-ion batteries, the hydrodynamics of flow batteries, and the mechanical dynamics equations of flywheels. The spatio-temporal correlation features of multi-source heterogeneous data are extracted through the Transformer architecture to generate a dynamic meta-knowledge base containing cross-scale non-linear interaction features. This model realizes the dynamic weight allocation of the coupling boundary of different physical fields through a multi-task learning framework with shared hidden layers, overcoming the modeling deviation of traditional single-physical-field models in scenarios of second-level transient response and hour-level cyclic decay.

[0146] In the dynamic parameter compensation stage, the Butler–Volmer electrochemical equation is embedded in the gating mechanism of the long short-term memory network. The phase change of the electrode and the diffusion characteristics of the electrolyte are separated through orthogonalization constraints to eliminate the cross-sensitivity between the state of charge and the state of health. A flow channel topology model of the flow battery is constructed by combining a graph neural network. Federated learning aggregates the operation data of multiple node pumps to generate a feedforward compensation coefficient matrix. The compensation residuals are spatio-temporally aligned to form a multi-dimensional state observation vector, providing accurate input features for multi-agent reinforcement learning and solving the problem of the lag response of traditional control strategies to non-linear dynamic characteristics.

[0147] At the level of closed-loop optimization control, the quantum-inspired optimization algorithm maps the power distribution threshold of the cooperative control strategy to the Hamiltonian constraint of the quantum annealing model, and dynamically updates the model boundary conditions through the harmonic spectrum features extracted by the deep residual network. The federated reinforcement learning framework uses the differential privacy mechanism to fuse multi-node real-time data, and realizes the lightweight migration of cloud policies to edge controllers through the imitation learning generative adversarial network. The execution results are injected back into the dynamic meta-knowledge base to trigger the recalibration of the parameters of the neural differential equation. Through the two-way feedback mechanism of the physical model and data-driven, this closed-loop architecture continuously corrects the dynamic mismatch of the multi-physical field coupling model, and realizes the adaptive optimization of the control strategy for non-linear characteristics.

Claims

1. An energy-saving optimization method for an electric energy storage device, characterized in that, Including: Construct a multi-physics coupling model of the energy storage device. By integrating the solid-liquid phase change kinetics equation of lithium-ion batteries, the fluid mechanics equation of flow batteries, and the rigid-flexible coupling equation of flywheel rotors through neural differential equations, generate a dynamic meta-knowledge base containing cross-time-scale nonlinear interaction characteristics; Based on the physical constraint conditions of the dynamic meta-knowledge base, use a physics-informed neural network to perform dynamic parameter compensation processing on the polarization voltage hysteresis effect and electrolyte flow resistance coupling effect of the energy storage device, and output a residual feature vector containing the decoupled state-of-charge characteristics; Input the residual feature vector into a multi-agent reinforcement learning framework to generate the hierarchical control parameters of the supercapacitor terminal voltage and the magnetic bearing damping compensation amount of the flywheel energy storage, forming a cooperative control strategy; According to the power distribution threshold of the cooperative control strategy, use a quantum-inspired optimization algorithm to transform the DC bus ripple suppression and harmonic distortion rate optimization objectives into the Hamiltonian constraint conditions of the quantum annealing model, and generate a converter modulation command; In the federated reinforcement learning framework, distill the cloud global optimization strategy to the edge controller through an imitation learning generative adversarial network, and at the same time inject the real-time operating condition data fed back by the edge controller into the multi-physics coupling model for parameter recalibration, forming a closed-loop energy efficiency enhancement loop across physical models, data-driven, and quantum optimization; The dynamic parameter compensation includes: On the lithium-ion battery side of the energy storage device, embed the Butler–Volmer electrochemical equation into a long short-term memory network to decouple the cross-sensitivity of the state-of-charge and health state in real time; On the flow battery side, construct a graph neural network to predict the flow channel pressure drop distribution, combine federated learning to aggregate the operation data of multi-node pump units, dynamically optimize the feedforward compensation coefficient matrix of the electrolyte flow rate, and use the compensation residual as the state observation input of the multi-agent reinforcement learning.

2. The energy-saving optimization method for the electrical energy storage device according to claim 1, wherein The construction of the multi-physics coupling model of the energy storage device includes: Integrate the solid-liquid phase change reaction kinetics equation of lithium-ion batteries, the fluid mechanics equation of flow batteries, and the rigid-flexible coupling equation of flywheel rotors using neural differential equations; Extract the spatio-temporal correlation features of multi-source heterogeneous data of the energy storage device through a Transformer architecture with spatio-temporal attention mechanisms, and generate a meta-knowledge base containing cross-scale nonlinear interaction relationships, which is used to provide physical constraint conditions for the physics-informed neural network.

3. The energy-saving optimization method for the electrical energy storage device according to claim 1, characterized in that Inputting the residual feature vector into a multi-agent reinforcement learning framework to generate the hierarchical control parameters of the supercapacitor terminal voltage and the magnetic bearing damping compensation amount of the flywheel energy storage, and forming a cooperative control strategy includes: Adopt a hierarchical attention mechanism for the supercapacitor terminal voltage control, and match the charge relaxation time constant through a sliding mode parameter dynamic adjustment strategy; Deploy a rotor deformation observer based on neural radiance fields on the flywheel energy storage side, and combine the proximal policy optimization algorithm to generate the magnetic bearing damping compensation amount, so that the quantum-inspired optimization algorithm receives the torque fluctuation suppression threshold as the modulation target constraint.

4. The energy-saving optimization method for the electrical energy storage device according to claim 1, characterized in that The multi-objective optimization according to the power distribution threshold of the cooperative control strategy includes: Transform the DC bus ripple suppression, harmonic distortion rate optimization, and switching loss minimization into the Hamiltonian of the quantum annealing model; Predict the harmonic spectrum distribution characteristics through a deep residual network, and feedback the distribution characteristics to the multi-physics field coupling model to update the boundary conditions of the neural differential equation.

5. The energy-saving optimization method for the electrical energy storage device according to claim 1, characterized in that The formation of a closed-loop energy efficiency enhancement loop integrating cross-physical models, data-driven methods, and quantum optimization includes: Based on the output of the multi-agent reinforcement learning framework, deploy the multi-agent proximal policy optimization algorithm in the cloud, and fuse the operation data of multi-node pump units aggregated by the federated learning through a differential privacy protection mechanism; On the edge side, use an imitation learning generative adversarial network to convert the cloud policy into lightweight control instructions, and inject the instruction execution results back into the generated dynamic meta-knowledge base to trigger the recalibration of the boundary condition parameters of the neural differential equation.

6. The energy-saving optimization method for the electrical energy storage device according to claim 1, wherein, The federated reinforcement learning framework includes: Based on the physical constraints of the dynamic meta-knowledge base, design a hierarchical reward mechanism, map the dynamic efficiency of the energy storage system to the depth of the potential energy well, and construct a long-term energy efficiency optimization goal; Generate a dynamic energy efficiency entropy index through a multi-entropy maximization controller, and this index is used to constrain the charge relaxation threshold of the supercapacitor power distribution interval in the generated cooperative control strategy.

7. The energy-saving optimization method for the electrical energy storage device according to claim 1, characterized in that It also includes: Embed an adversarial curriculum generator within the federated learning framework to synthesize training scenarios based on the critical operating condition data stream; The federated learning framework performs differential privacy processing on the operation data of multi-node pump units to generate differentially private protected data; Evaluate the distribution difference degree between the synthesized scenario and the differentially private protected data through a graph attention network, and dynamically adjust the training curriculum difficulty coefficient of the multi-agent reinforcement learning framework.

8. The energy-saving optimization method of the electrical energy storage device according to claim 4, characterized in that, It also includes: Construct an unsteady Pareto front search engine, and dynamically allocate multi-objective optimization weights according to the dynamic energy efficiency entropy index and harmonic spectrum distribution characteristics; Input the weight sequence into the quantum-inspired optimization algorithm to update the Hamiltonian constraint conditions of the quantum annealing model to achieve real-time alignment of the converter modulation strategy and the dynamic energy efficiency optimization goal.

9. The energy-saving optimization method of the electrical energy storage device according to claim 1, characterized in that The closed-loop energy efficiency enhancement loop includes: Inject the synthesized critical operating condition data stream back into the dynamic meta-knowledge base to trigger the reparameterization of the fluid dynamics equation of the flow battery in the neural differential equation; Constraining the frequency response interval of the flywheel energy storage magnetic bearing damping compensation amount in the generated cooperative control strategy through the hierarchical reward mechanism dynamic energy efficiency entropy index; Synchronously feedback the updated dynamic meta-knowledge base parameters to the physical information neural network and the quantum optimization algorithm to form a closed-loop enhancement loop for cross-model parameter compensation, control strategy generation, and modulation optimization.

Images