Energy scheduling and decentralized transaction management system based on battery energy storage system

Abstract

The application relates to the technical field of power systems and their automation, and particularly discloses an energy scheduling and decentralized transaction management system based on a battery energy storage system. The health states of battery modules are calculated and updated by using a Bayesian algorithm through the collection of battery multi-modal health data by sensors; system-level virtual state of charge balancing is realized based on a chemical neutral virtual battery abstraction layer; the energy surplus or deficit of each node is predicted by a causal state space model; the optimal energy supply path is calculated by a multi-hop energy balance controller for the deficit nodes; and the transaction price is generated by a personalized energy token pricing engine in combination with the battery health state, operating conditions and the like, and finally automatically settled through a blockchain smart contract. The application can deeply perceive the health and chemical properties of heterogeneous batteries, improve the asset life and system safety, and enhance the economy and reliability of a distributed energy storage network.

Description

Technical Field

[0001] This invention belongs to the field of power system and its automation technology, specifically relating to an energy dispatch and decentralized trading management system based on a battery energy storage system. Background Technology

[0002] With the rapid development of renewable energy and distributed energy storage systems, battery energy storage systems have played a significant role in improving grid flexibility and user-side energy autonomy. However, existing energy storage systems often operate independently, are diverse in type, and are geographically dispersed, making it difficult to achieve collaborative optimization and resource sharing. Traditional centralized dispatching carries the risk of single points of failure and is ill-suited for multi-entity, highly dynamic distributed scenarios.

[0003] Current energy trading solutions based on decentralized architecture still have significant shortcomings: First, they lack real-time awareness of battery health status and degradation effects, and pricing does not take into account battery loss and safety constraints, which can easily affect battery life; second, energy dispatch path selection is relatively simple, without comprehensively considering transmission loss, equipment efficiency, and path reliability, making it difficult to achieve efficient and reliable energy transmission; third, the system has poor compatibility with batteries of different chemical systems and lacks a unified coordination mechanism, which limits the collaborative operation of heterogeneous energy storage.

[0004] Therefore, existing technologies struggle to achieve efficient, reliable, and highly compatible distributed energy storage collaborative scheduling and trading while ensuring battery safety and extending asset lifespan, thus limiting the overall economic efficiency and reliability of energy storage networks. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide an energy dispatch and decentralized transaction management system based on battery energy storage system, which can effectively solve the problems in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: an energy dispatch and decentralized transaction management system based on a battery energy storage system, the system comprising the following components:

[0007] A distributed battery energy storage system cluster consists of multiple physically dispersed, electrically isolated, and heterogeneous independent battery energy storage systems. Each battery energy storage system is equipped with a local controller, sensor array, and communication interface.

[0008] The active health inference module is used to inject diagnostic pulses into the battery energy storage system and collect multimodal sensor data. It evaluates the health status of each battery module based on the Bayesian algorithm and uploads the results to the blockchain network.

[0009] The chemically neutral virtual battery abstraction layer is used to establish chemical transfer functions for battery modules with different chemical systems, realize the mapping from power request to estimated efficiency, temperature rise and incremental wear, and perform chemically sensed virtual state of charge balancing and coordinated scheduling.

[0010] A deterministic multi-hop energy balance controller is used to predict node energy deficits based on a causal state-space model, calculate composite transmission costs including power loss costs, battery wear costs, and trust penalties, and schedule the optimal multi-hop energy transmission path through a constrained optimization model.

[0011] A personalized energy token pricing engine is used to dynamically generate personalized transaction prices based on battery wear and tear costs, system stability requirements, and on-chain reputation, and automatically execute settlement through blockchain smart contracts.

[0012] The dynamic safety reconfiguration module is used to reconfigure the power topology through solid-state switches when a battery pre-runaway signal is detected, thereby achieving fault isolation and load transfer.

[0013] The user-programmable policy interface is used to receive and execute user-defined charging and discharging policies, reserve capacity rules, energy sources, and trading conditions.

[0014] Furthermore, the health inference process performed by the active health inference module specifically involves: periodically injecting diagnostic pulses with a duration of 1 to 5 seconds and an amplitude less than 10% of the rated current; simultaneously monitoring the voltage transient response, effective pulse impedance, battery surface and core temperature gradient, acoustic emission spectrum characteristics, micro-pressure or micro-strain drift, operating duty cycle, cumulative discharge depth, throughput, and HVAC system status during the pulse; and constructing a feature vector containing the above features.

[0015] The posterior probability of each battery module being in a healthy, suspicious, or abnormal state is calculated based on Bayesian rules. The Bayesian rules are expressed as follows: ,in This is a prior probability based on historical data and the module's age. Let be the likelihood probability density of the observed eigenvectors under a given state. To observe the marginal probabilities of the feature vectors, a normalization constant is used; the posterior probabilities are normalized, and an evidence-weighted health index is calculated. , Preset weights for each state; classify modules based on the BEWHI index: A value ≥0.7 is considered a healthy state, and 0.4 ≤ A value <0.7 is considered suspicious. If the value is less than 0.4, it is considered an abnormal state, and the corresponding degraded operation, partial rebalancing, isolation, or maintenance work order instruction is triggered.

[0016] Furthermore, the cross-chemical scheduling optimization process implemented by the chemically neutral virtual battery abstraction layer is as follows: for battery modules with different chemical systems such as lithium iron phosphate, nickel manganese cobalt, sodium ion, and secondary batteries, their chemical transfer functions are established respectively. The function maps the requested power Preq to a triplet ( , , ), representing the estimated efficiency, estimated temperature rise, and estimated incremental wear, respectively;

[0017] The optimizer solves the objective function Minimize:∑( + + ),in, To estimate the cost of electricity loss, The wear shadow price cost is calculated based on incremental wear. To mitigate the thermal risk penalty costs associated with the estimated temperature rise; the optimization process simultaneously satisfies the voltage window constraints, maximum charge / discharge current constraints, and overall system response time constraints of each module; by dynamically pairing battery modules with fast response characteristics with battery modules with high energy density but slower response, their power output is coordinated in the form of phase difference to suppress DC bus voltage deviation.

[0018] Furthermore, the composite unit energy transfer cost for path selection in a deterministic multi-hop energy balance controller. Cost of power loss Wear shadow price cost based on incremental wear calculation and trust penalties The sum of the three constitutes, that is ;

[0019] Cost of power loss The path resistance loss and converter efficiency loss are monetized, and the calculation formula is as follows: ,in For reference energy prices, Energy is lost due to resistance. Energy loss due to inverter; cost of donor battery wear and tear. It reflects the marginal degradation caused by each kilowatt-hour of energy delivered, and its calculation is based on baseline throughput cost and operating point stress multiplier determined by temperature, charge / discharge rate, state of charge and depth of discharge;

[0020] Trust Punishment The calculation depends on path trust score ,in Let h be the path hop count, γ be the hop count decay factor, d be the total path distance, and λ be the distance decay constant. The trust penalty function is: ,in, This is the preset maximum penalty amount.

[0021] Furthermore, the personalized energy token pricing engine generates transaction prices. Settlement price from the underlying market Battery Wear Shadow Price (BWSP), Stability Reserve Premium (SRP), and Reputation Discount Joint decision, that is ;

[0022] Battery wear shadow price (BWSP) is determined by the donor battery wear cost. Derived and superimposed with real-time thermal safety margin penalty; Stability Reserve Premium (SRP) is the pricing of the scarcity of stability attributes such as the current required reserve capacity, decreasing reserve capacity, and virtual inertia contribution of the microgrid, and its value originates from the Lagrangian dual variable or equivalent scarcity function of the system-level reserve constraint; Reputation Discount It is a function of the reliability score s on the donor node chain. ,and , This is the preset maximum discount limit.

[0023] Furthermore, upon receiving a pre-runaway signal, the dynamic safety reconfiguration module executes a load redistribution strategy. This strategy calculates the optimal allocation scheme for the original load of the isolated module based on the proximity principle, the current load rate and health status index of each healthy module, and completes the dynamic reconfiguration of the power topology and seamless load switching within milliseconds by controlling the solid-state switch network.

[0024] Furthermore, the asset owner policies received by the user-programmable policy interface include: charge / discharge policies, used to set charging or discharging during specific electricity price periods; reserve capacity rules, used to set the minimum reserve capacity percentage that must always be maintained; acceptable energy source rules, used to limit the energy type of the trading object; and trading condition rules, used to set the minimum selling price, the maximum purchasing price, or the minimum credit score threshold of the trading object.

[0025] In summary, this application includes at least one of the following beneficial technical effects:

[0026] (1): By introducing active health inference and chemically neutral virtual battery abstraction layer, we have achieved deep perception and unified scheduling of the health status and chemical characteristics of heterogeneous battery energy storage system. The wear cost and thermal safety constraints of the battery body are embedded into the scheduling and pricing core, which effectively avoids accelerated aging caused by overuse and improves asset life and system safety.

[0027] (2): By adopting a deterministic multi-hop energy balance controller and a composite cost optimization model, when carrying out cross-node energy scheduling, multiple factors such as line ohmic loss, converter efficiency, multi-hop trust decay and node reliability are comprehensively considered, achieving high energy efficiency and high reliability of optimal path selection and pre-scheduling, which significantly improves the overall economy and operational reliability of distributed energy storage networks.

[0028] (3): A closed-loop control and transaction framework integrating health inference, wear pricing, stability premium and on-chain reputation mechanism was constructed. It supports user-defined strategies and realizes truly decentralized, autonomous, self-healing and economically incentive-compatible energy scheduling and transaction. It breaks through the single point of failure risk of traditional centralized architecture and the bottlenecks of existing decentralized solutions in terms of coordination, security and interoperability. Attached Figure Description

[0029] Figure 1 This is a schematic diagram of the overall technical solution architecture of the energy dispatch and decentralized transaction management system based on battery energy storage system proposed in this invention.

[0030] Figure 2 This is a schematic diagram of the core principle framework of the deterministic multi-hop energy balance controller in this invention;

[0031] Figure 3 This is a logical flow diagram of the personalized energy token pricing engine and blockchain transaction execution in this invention. Detailed Implementation

[0032] To further illustrate the technical means and effects adopted by the present invention to achieve the intended purpose, the specific embodiments according to the present invention will be described in detail below with reference to the accompanying drawings and preferred embodiments.

[0033] In a regional energy network comprised of multiple industrial and commercial parks, residential communities, and independent microgrids, the energy dispatch and decentralized trading management system based on battery storage systems, as described in this invention, is deployed. This network contains numerous physically distributed, electrically isolated, and independently owned battery storage systems, with capacities ranging from tens of kilowatt-hours to several megawatt-hours, and electrochemical systems encompassing various types such as lithium iron phosphate, nickel-manganese-cobalt ternary, sodium-ion, and lead-acid rechargeable batteries. Each battery storage system is equipped with a local controller, sensor array, and communication interface, collectively forming a distributed battery storage system cluster. The core objective of the system operation is to achieve efficient, secure, and reliable decentralized energy sharing and trading across ownership while meeting the load and reserve requirements of each node, thereby maximizing the overall network's economy and resilience.

[0034] Combined with appendix Figure 1-3The overall architecture of this system consists of a distributed battery energy storage system cluster, an active health assessment module, a chemically neutral virtual battery abstraction layer, a deterministic multi-hop energy balance controller, a personalized energy token pricing engine, a dynamic security reconfiguration module, and a user-programmable strategy interface. Each module interacts with the blockchain ledger via a high-speed communication network for data exchange and command synchronization, forming a closed-loop autonomous system integrating perception, decision-making, execution, and settlement.

[0035] The proactive health assessment module is the primary component for achieving safe and economical battery management. Its operation mainly includes the following sub-steps:

[0036] Step 1: Inject low-amplitude diagnostic pulses

[0037] This module periodically injects low-amplitude diagnostic pulses into designated battery modules in each battery energy storage system. The duration of the diagnostic pulse is controlled between 1 and 5 seconds, and the pulse current amplitude is less than 10% of the module's rated charge and discharge current, in order to avoid interfering with the normal operation and lifespan of the battery.

[0038] Step 2: Synchronously acquire multimodal sensor data

[0039] Within the same time window of injecting diagnostic pulses, the module synchronously acquires multimodal sensing data streams through the sensor array under unified timing control. Specifically, the local controller sends synchronization trigger signals to each relevant sensor, or assigns a unified high-precision timestamp to each sensor's data, ensuring strict time alignment of the acquired multi-channel data. The acquired data includes:

[0040] The voltage transient response curve recorded by the voltage sensor is used for subsequent calculation of the effective pulse impedance;

[0041] Temperature gradient changes are recorded by temperature sensors distributed on the battery surface and core.

[0042] Acoustic emission spectral characteristics captured by acoustic emission sensors, caused by changes in the microstructure inside the battery;

[0043] Micro-strain sensors monitor micro-pressure or micro-strain drift signals in the battery casing or internal structure.

[0044] Step 3: In addition, this module continuously records and maintains historical operating data for each battery module. This data is automatically calculated, generated, and stored by the local controller based on sensor readings and control signals, and will serve as a long-term reference benchmark for subsequent health status assessments. The main recorded data includes: operating duty cycle characterizing the battery's working mode; cumulative depth of discharge reflecting battery usage intensity; total throughput characterizing the battery's cumulative workload; and the associated operating status of the HVAC system used for correlation analysis of the battery's thermal history.

[0045] In summary, the proactive health inference module periodically injects diagnostic pulses with specific parameters and simultaneously collects real-time sensor data from multiple dimensions, including electrical, thermal, acoustic, and mechanical data. Combined with the battery's historical operating records, this provides a complete and diverse data foundation for subsequent Bayesian inference-based health status calculations. This step is the starting point of the entire state perception process, ensuring that the input information for subsequent health inference is sufficient and synchronized.

[0046] The proactive health assessment module is the primary component for achieving safe and economical battery management. The implementation process of this module can be clearly divided into the following three consecutive steps:

[0047] Inject parameterized low-amplitude diagnostic pulse steps:

[0048] According to a preset scheduling strategy, this module periodically injects low-amplitude diagnostic pulses into designated battery modules in each battery energy storage system. The parameters of the diagnostic pulses are specifically configured as follows: their duration is controlled between 1 and 5 seconds, and the pulse current amplitude is strictly limited to less than 10% of the module's rated charge / discharge current. The direct purpose of this parameter design is to effectively excite the battery to generate observable response signals while minimizing interference with the normal operation and long-term lifespan of the battery caused by diagnostic behavior.

[0049] Steps for synchronously acquiring multimodal real-time sensor data:

[0050] Within the same time window of the injected diagnostic pulse, the module synchronously acquires multi-dimensional physical responses of the battery through an integrated sensor array. To achieve data time alignment, the local controller sends a unified trigger signal to each sensor or assigns a high-precision time stamp to the acquired data. The specific real-time data stream acquired includes:

[0051] Electrical response: The voltage transient response curve recorded by the voltage sensor is used for subsequent calculation of the battery's effective pulse impedance;

[0052] Thermal response: Temperature gradient change data recorded by temperature sensors distributed on the surface and core of the battery;

[0053] Acoustic response: The acoustic emission spectrum characteristics captured by acoustic emission sensors, caused by changes in the microstructure inside the battery (such as lithium dendrite growth and separator deformation).

[0054] Mechanical response: Micro-pressure or micro-strain drift signals monitored by micro-strain sensors in the battery casing or internal structure.

[0055] This step aims to obtain a "panoramic" instantaneous snapshot of the battery's state under specific stimuli.

[0056] Steps for recording and maintaining historical operational data:

[0057] In addition to real-time sensor data, this module continuously records and maintains historical operating data for each battery module. This data is automatically calculated, generated, and stored by the local controller based on continuous monitoring signals. The core historical parameters recorded include:

[0058] The duty cycle that characterizes the battery's operating mode;

[0059] The cumulative depth of discharge reflects the intensity of battery use and the depth of cycle time;

[0060] Total throughput, representing the cumulative workload of the battery;

[0061] Used to correlate the battery thermal management history with the associated operating status of the heating, ventilation, and air conditioning (HVAC) system.

[0062] These historical data form a baseline reference for assessing the long-term degradation trend and health status of batteries.

[0063] After collecting real-time and historical data, the proactive health inference module executes the following core algorithm steps to calculate the quantitative health index of each battery module and complete the status classification:

[0064] Step A: First, based on the collected synchronous multimodal real-time data (electrical, thermal, acoustic, and mechanical responses) and recorded historical operational data, the module performs preprocessing such as data cleaning, formatting, and normalization to convert the raw data with different physical dimensions and time series into a unified numerical format. Then, these preprocessed multidimensional data are concatenated and combined in a predetermined order to construct a fixed-dimensional comprehensive feature vector that can be used for probabilistic model calculations. This feature vector serves as input evidence for subsequent Bayesian probabilistic inference.

[0065] Step B: Determine health status and calculate prior probabilities

[0066] The system pre-determines three discrete health states for each battery module: healthy, suspicious, and abnormal. Before each inference, the module calculates the prior probability P(State) of each state based on the individual information and group knowledge of the battery.

[0067] Specifically, the prior probability can be calculated using a combination of one or more of the following methods:

[0068] Reliability models based on runtime and loop count: For example, using reliability models such as the Weibull distribution, the basic probability of the module being in an "abnormal" state can be estimated based on the module's cumulative runtime and loop count.

[0069] Interpolation based on historical performance degradation: Based on the historical capacity or internal resistance degradation curve of the module since it was manufactured, assess its deviation from the health threshold and map it to the initial probability of a "suspicious" or "abnormal" state.

[0070] Proportional allocation based on group failure statistics: Referring to historical failure statistics of battery groups of the same model under similar operating time and number of cycles, an empirical initial probability proportion is directly assigned to the "healthy", "suspicious", and "abnormal" states.

[0071] This step, by quantifying the individual aging process and group risk of batteries, provides an initial judgment based on objective history for subsequent Bayesian inference.

[0072] Step C: Perform Bayesian inference. This step is the core computational process for health inference, and its goal is to update the understanding of battery health status based on new evidence (feature vectors).

[0073] C1. Model Preparation: The system relies on a pre-trained likelihood probability density function. This function is obtained by collecting a large amount of multimodal feature vector data of the same type of battery modules under known health conditions (e.g., through laboratory calibration or long-term field operation verification) as a training set. Using parametric estimation (e.g., Gaussian mixture model) or non-parametric density estimation methods, corresponding feature vector probability distribution models are established for the three states of "healthy", "suspicious", and "abnormal". This model expresses the probability of observing a certain feature vector under a given state, i.e., the likelihood probability density function.

[0074] C2. Calculate the likelihood probability: For the battery module to be evaluated, the system calls the model trained above. Input the comprehensive feature vector constructed in step A into the model, calculate the conditional probability of observing the current feature vector under the assumption that the battery is in a certain health state (such as "healthy"), that is, obtain P(Evidence|State).

[0075] C3. Calculate the posterior probability: Apply Bayes' rule, and combine the prior probability P(State) obtained in step B with the likelihood probability density calculated in step C2. Multiply, then divide the result by the marginal probability of observing the eigenvector. The posterior probabilities of the battery module being in each health state are obtained. .

[0076] C4. Normalization: Normalize the obtained posterior probabilities to ensure that the sum of the posterior probabilities of the three states "healthy", "suspicious", and "abnormal" is 1, thereby obtaining a standardized probability distribution.

[0077] Step D: Calculate the Evidence-Weighted Health Index (BEWHI)

[0078] To obtain an intuitive and quantifiable health assessment indicator, the module calculates the Evidence-Weighted Health Index (BEWHI). The calculation formula is as follows: , Preset weights for each state, This represents the posterior probability of the state.

[0079] The system pre-determines weights for each state: healthy states have the highest weight (positive values), followed by suspicious states (lower positive values), and abnormal states have negative weights. These weights are pre-configured based on the impact of different battery health states on system safety and economy. This design allows the BEWHI index to comprehensively reflect the probability distribution and impose clear penalties on negative health states.

[0080] Step E: Status Classification and Handling

[0081] Based on the calculated BEWHI index, the module performs automatic classification and linkage control:

[0082] If BEWHI ≥ 0.7, the module is considered to be in a healthy state. This module can handle all normal tasks and some backup tasks.

[0083] If 0.4 ≤ BEWHI < 0.7, the system is considered suspicious. It will automatically implement a downgraded operating strategy, such as limiting the maximum charge / discharge rate, and may trigger a local rebalancing command.

[0084] If BEWHI < 0.4, it is considered an abnormal state. The system immediately generates a maintenance work order and notifies the subsequent dynamic safety reconfiguration module to prepare for electrical isolation from the available resource pool.

[0085] Step F: Data on-chain storage and evidence preservation

[0086] All inference results, including the Health Index (BEWHI), final classification results, and feature vector data used for inference, are encrypted and uploaded to the blockchain network. This creates an immutable and traceable digital health profile for the battery. This reliable profile provides an authoritative data foundation for subsequent decisions regarding the "Personalized Energy Token Pricing Engine" and the "Deterministic Multi-hop Energy Balance Controller."

[0087] The chemically neutral virtual battery abstraction layer and the deterministic multi-hop energy balance controller in this application work together to form the core of achieving unified scheduling and optimal energy allocation across nodes for heterogeneous batteries. Its implementation process can be clearly divided into the following stages:

[0088] Phase 1: Construction and Operation of the Abstraction Layer of a Chemically Neutral Virtual Battery

[0089] Step M1: Establish the chemical transfer function of the battery module

[0090] The chemically neutral virtual battery abstraction layer receives health status data from the active health probing module and integrates the factory calibration parameters of each battery module, such as chemical system, rated capacity, voltage window, and internal resistance characteristic curve. Based on this data, this layer establishes a unique chemical transfer function for each type of battery module with a specific chemical system.

[0091] Specifically, chemical transfer functions can be constructed and calibrated through one or a combination of the following methods: (1) parameterization and simplified derivation based on battery electrochemical mechanism models (such as equivalent circuit models and single-particle models); (2) data-driven modeling using historical test data of the battery at different power, temperature, and state of charge, through machine learning methods (such as neural networks and Gaussian process regression). The modeling goal is to establish a mapping relationship from requested power to output triplet.

[0092] Functional purpose: As a mathematical model, the core function of this function is to map the power request value input from the outside to the internal response of the battery module at this operating point.

[0093] Output: For a given requested power, the function outputs a triplet prediction, including:

[0094] Estimated efficiency ( ): Energy conversion efficiency at this power point.

[0095] Forecast temperature rise ( ): The battery temperature rises as a result of executing this power request.

[0096] Predicted incremental wear ( The micro-loss of battery life caused by executing this request can be reflected as an equivalent measure of capacity decay or internal resistance increase.

[0097] Step M2: Perform chemically sensed virtual state of charge equilibration

[0098] Based on the model established in step M1 and the real-time health status, this layer performs "chemical sensing" virtual balancing that surpasses traditional power balancing.

[0099] Core operation: It adaptively and differentially sets real-time charge and discharge operation boundaries (such as voltage window, maximum allowable current) for each available battery module in the network.

[0100] Decision-making basis: When setting boundaries, both the chemical characteristics and current health status of the battery are considered. For example, the maximum charging current boundary of a healthy but slow-responding lithium iron phosphate battery may be set lower than that of a healthy and fast-responding ternary lithium battery; while the allowable state of charge range of a sodium-ion battery whose health status is deemed "questionable" will be actively narrowed to avoid operation in the high and low voltage range where aging is likely.

[0101] Step M3: Achieve phase coordination of heterogeneous modules through optimization.

[0102] To achieve optimal control of the overall power response of the system, this layer solves the system-level optimization problem.

[0103] Optimization objective: Minimize the total cost, which is the cost of electricity loss (based on...). Wear and tear shadow price cost (based on) ) and thermal risk penalty costs (based on The weighted sum of the three.

[0104] Optimization constraints: The solution process must strictly meet the constraints such as the real-time voltage window of each module, the maximum charging and discharging current determined by the health status, and the overall system response time.

[0105] Coordination Effect: Through optimization, the system can dynamically pair battery modules with different characteristics. For example, modules with faster response times can act first to smooth out power surges, followed by modules with higher energy density for a smooth transition. This "phase wave" style coordination control effectively suppresses DC bus voltage deviations and achieves optimal power matching between batteries with different chemical systems.

[0106] Phase 1 Summary: The above three steps enable the system to quantitatively evaluate the real-time capabilities and cost of batteries with different chemistry and health states in a unified virtual layer, and assign them appropriate working roles and boundaries, providing accurate "cost" and "capability" inputs for cross-node scheduling.

[0107] Phase Two: Scheduling Decisions of Deterministic Multi-Hop Energy Balance Controller

[0108] Step N1: Predict energy supply and demand based on a causal state-space model

[0109] The deterministic multi-hop energy balance controller first integrates multi-source data, including real-time load of each node, status of neighboring nodes, ultra-short-term predicted output of renewable energy, and user energy consumption patterns based on historical analysis. Using a causal state-space model, the controller predicts the energy surplus or deficit of each node in the distributed network in a specific future period. This prediction considers not only the total energy but also the temporal characteristics of power changes.

[0110] The causal state-space model adopts the S4 (Structured State Space Sequence Model) architecture, and its continuous-time state equation is dx / dt=Ax+Bu,y=Cx, where A∈ N×N is the diagonalized state matrix, B∈ N×1, C∈ 1×N represents the learnable parameters; the model input consists of the node load, photovoltaic output, and electricity price sequence for the past 24 hours, and the output is the predicted net energy demand every 15 minutes for the next 4 hours; the model is trained end-to-end on a historical dataset containing at least 1000 typical operating days by minimizing the mean square error between the predicted and measured values.

[0111] Step N2: Calculate the combined transmission cost of potential power supply paths

[0112] For nodes predicted to have energy deficits, the controller calculates the composite unit energy transmission cost for all possible power supply nodes and multi-hop transmission paths. This cost serves as a core indicator for economic optimization. It consists of three parts:

[0113] Cost of power loss ( ): Resistance losses along the monetization path and efficiency losses across all converters along the path. Calculations depend on line resistance, converter efficiency curves, and reference energy prices.

[0114] Donor battery wear cost ): Reflects the marginal lifetime degradation cost of the donor battery due to energy transfer. Its calculation is based on the baseline cost associated with the battery chemistry and is corrected by an "operating point stress multiplier" according to the battery's real-time temperature, charge / discharge rate, state of charge, and depth of discharge at the time of the transfer.

[0115] Trust penalty cost ( ): Quantify the reliability risk of a path. Its calculation is based on a path trust score ( This score is the product of the historical reliability scores of each node on the path, and it decays exponentially with the number of hops and physical distance. The trust penalty cost is the preset maximum penalty amount multiplied by (1- The product of ).

[0116] Step N3: Solve the optimization problem and generate pre-scheduled instructions.

[0117] The controller model predictive control optimization problem is to minimize the total scheduling cost (i.e., total transmission energy × 100%) over the entire prediction time domain, while satisfying hard constraints such as all converter power limits, DC bus voltage safety windows, and the state of charge and temperature boundaries of each node. By solving this problem, the controller can determine one or more globally optimal energy transmission paths, transmission power profiles, and precise start-up and shutdown times before the deficit actually occurs, and generate pre-scheduled energy package instructions with timestamps.

[0118] Phase Two Summary: This phase achieves automated, economically optimal planning and pre-scheduling of cross-node, multi-hop energy transmission paths through forward-looking prediction and a composite economic model that includes real losses (electrical losses, battery wear) and virtual risks (trust costs).

[0119] After completing the composite cost calculation for all potential energy supply paths, the deterministic multi-hop energy balance controller enters the final optimization decision and command generation stage. This process aims to transform economic calculations into executable control commands, and its core can be decomposed into the following three steps:

[0120] Step P1: Construct a model predictive control optimization problem. The controller is based on the composite unit energy transmission cost of each path calculated in step N2. ), and construct a model predictive control optimization problem covering the entire prediction time domain.

[0121] Optimization objective: Minimize the total compound cost of all scheduling actions during the forecast period.

[0122] Optimization variables: The decision variables for this problem specifically include whether to select a specific multi-hop transmission path for the red-byte point, when to start and end energy transmission, the power level for transmission, and the duration of this transmission. These variables collectively determine the specific scheduling scheme.

[0123] Step P2: Define and apply hard constraints to the system. Solving the optimization problem is not unconditional; it must strictly satisfy a series of hard constraints reflecting the safety and operational limits of the physical system. These mainly include:

[0124] Equipment capacity constraints: All converters involved must not exceed their maximum permissible power, and their operating voltage must also be within the safety window.

[0125] Network stability constraint: The voltage of the DC bus must be maintained within the safe operating window.

[0126] Scheduling feasibility constraints: Energy transmission scheduling must meet the time window requirements to ensure completion before a deficit occurs.

[0127] Battery safety constraints: The state of charge of all donor and recipient battery nodes must be maintained within the preset safety boundaries throughout the entire scheduling process, and overcharging or over-discharging is prohibited.

[0128] Thermal safety constraints: The temperature of all battery modules must not exceed their safety threshold at any time.

[0129] Step P3: Solve the optimization and generate executable instructions. The controller obtains the globally optimal scheduling scheme by solving the above-mentioned constrained optimization problem (usually using standard optimization algorithms such as linear programming, quadratic programming or mixed integer programming).

[0130] Output results: The solution results clearly identify one or more globally optimal energy transmission paths, the precise power-time curves (power profiles) for each path, and the precise start and end times of transmission.

[0131] Instruction Generation: Based on this result, the controller generates a set of pre-scheduled energy package instructions with millisecond-level timestamps. These instructions are immediately sent to the local controllers of the relevant nodes, guiding them to prepare for execution. Simultaneously, the transaction intention corresponding to this scheduling instruction (including path, energy, time, participants, etc.) is created as a record and synchronously submitted to the blockchain network for notarization, providing a contractual basis for the subsequent automatic settlement of smart contracts.

[0132] The personalized energy token pricing engine is responsible for generating the final transaction price for each pre-scheduled energy package to be executed and driving subsequent automated settlement. Its pricing mechanism and settlement process can be broken down into the following steps:

[0133] Step Q1: Gather and confirm core pricing components

[0134] Just before the pre-scheduled energy pack instruction is executed, the pricing engine starts and gathers all the core components needed to calculate the final price. These components and their sources are as follows:

[0135] Underlying market clearing price ( This price reflects the universal time value of electricity. It can originate from the real-time clearing price of an external regional energy market, or from an internal benchmark price derived from matching and clearing based on current supply and demand conditions within the system.

[0136] Battery Wear Shadow Price (BWSP): This price is directly related to the lifetime loss and thermal risk borne by the donor battery due to this energy transfer. Its calculation is based on the donor battery wear cost provided by a deterministic multi-hop energy balancer. This is combined with a thermal safety margin penalty calculated based on the degree to which the real-time temperature of the donor battery approaches its safety limit, thereby achieving full economic compensation for the loss of the battery itself.

[0137] Stability Reserve Premium (SRP): This premium is the price paid for the scarcity of ancillary services required to maintain frequency and voltage stability in a microgrid. Its value stems from the real-time demand intensity at the system level for stability constraints such as rapidly increasing / decreasing reserve capacity and virtual inertia, and can typically be quantified by the Lagrangian dual variables of these constraints or an equivalent scarcity function. The more strained the system stability, the higher the premium.

[0138] Reputation discount ( This is an incentive given to highly reliable transaction nodes. The discount amount is a function of the reliability score(s) that the donor node updates in real time on the blockchain, i.e. The higher the rating, the larger the discount, but there is a preset upper limit D_max to prevent abuse.

[0139] Step Q2: Synthesize personalized transaction prices

[0140] The pricing engine dynamically combines the four core components mentioned above, according to the formula. The final personalized price for this energy transaction is calculated. This price comprehensively reflects the market value of electricity, the actual cost of physical losses, the service value of system stability, and the credibility of the transacting parties.

[0141] Step Q3: Deploy smart contracts and drive automatic settlement

[0142] Once the price is determined, the pricing engine encapsulates key information such as the final personalized price, transaction energy, identities of the participating parties, and pre-scheduled timestamps to form a standardized digital transaction contract, which is then deployed to the blockchain network.

[0143] Settlement Trigger: When the time point set in the pre-scheduling instruction arrives, and the local controllers of the relevant nodes confirm to the blockchain that the energy transfer has been completed precisely as instructed, the smart contract deployed on the chain will be automatically triggered.

[0144] Automatic execution: The smart contract automatically completes the transfer and settlement of tokens from the energy recipient to the donor according to the contract terms.

[0145] Default Handling: If either party (such as the donor failing to transmit electricity as agreed, or the recipient failing to confirm receipt) fails to fulfill its obligations, the smart contract will automatically execute the pre-set penalty clauses (such as deducting the deposit). The entire process requires no intervention from any centralized institution.

[0146] In summary, the personalized energy token pricing engine employs a dynamic pricing model that internalizes multi-dimensional costs and makes incentive mechanisms explicit. This model precisely integrates factors that are difficult to price in traditional markets, such as battery wear, system stability, and transaction credit, into the price of each transaction. This not only ensures the fairness and economic incentive compatibility of transactions, but also, through automated and trustless settlement achieved via blockchain smart contracts, constitutes a crucial link in the entire system's value loop, from scheduling decisions to value closure, thus guaranteeing the executability and ultimate credibility of decentralized transactions.

[0147] The dynamic safety reconfiguration module serves as the last line of defense for the system in response to sudden battery failures. Its core function is to achieve fault isolation and system self-healing within milliseconds. The specific workflow can be broken down into the following four consecutive steps:

[0148] Step R1: Fault warning signal reception and triggering

[0149] During system operation (including before energy transfer or during normal operation), if the proactive health detection module detects pre-runaway signals such as abnormal thermal gradients, impedance abrupt changes, or pressure drift in a battery module in real time, it will immediately send an alarm to the dynamic safety reconfiguration module. The module will then be triggered and enter emergency response mode.

[0150] Step R2: Millisecond-level rapid fault isolation

[0151] Upon receiving the signal, the module immediately controls the solid-state switch network deployed at the output of the target high-risk battery module, driving the relevant solid-state switches to disconnect within milliseconds (e.g., less than 10 milliseconds), thereby completely physically isolating the faulty or high-risk module from the main electrical circuit. The core purpose of this step is to prevent the fault from further developing or spreading, ensuring that the fault is contained to a minimum.

[0152] Step R3: Calculate the load redistribution scheme

[0153] While isolating the faulty module, the module calculates in real time, according to the preset load redistribution strategy, how the load originally borne by the isolated module should be transferred to other healthy battery modules in the network.

[0154] Strategy Basis: The calculation process typically considers the following factors:

[0155] Proximity principle: Prioritize adjacent modules that are close in electrical distance and have low connection loss.

[0156] Current load rate: Assess the current power output capacity and margin of each health module.

[0157] Health Status Index: Referencing the latest Evidence Weighted Health Index (BEWHI) for each module, prioritize allocating the load to modules with better health status.

[0158] By taking into account the above factors, the module calculates the optimal or near-optimal load transfer scheme and determines the additional power share that each target healthy module should bear.

[0159] Step R4: Dynamically reconfigure the topology and maintain power output

[0160] Based on the allocation scheme calculated in step R3, the module dynamically and quickly reconstructs the local power topology by controlling the closing or state switching of other related switches in the solid-state switch network. This operation seamlessly and smoothly transfers the load of the original faulty module to one or more designated healthy battery modules. Thus, while successfully isolating the fault source, it ensures the continuity and stability of the overall system's external power output, achieving a self-healing effect of "fault isolation without power interruption".

[0161] The user-programmable policy interface provides asset owners with autonomous control over the scheduling and trading activities of their battery energy storage systems. Its workflow can be summarized in the following three main steps:

[0162] Step S1: Strategy Setting and Input

[0163] Asset owners can use this interface to set personalized operating strategies for their battery systems through configuration files, graphical rule editing interfaces, or API calls. These strategies mainly cover the following types and can be configured in combination:

[0164] Charging and discharging strategy: Set charging and discharging behavior based on time or electricity price signals. For example, set "charge only during off-peak hours when the electricity price is below 0.4 yuan / kWh, and discharge during peak hours when the electricity price is above 0.8 yuan / kWh".

[0165] Backup capacity rules: Define the reserve energy that the system must maintain. For example, it may require that "the state of charge (SOC) of the battery must never fall below 20% of the total capacity at any time to ensure emergency power supply capability."

[0166] Acceptable energy sources: Limit the type of energy available to trading partners. For example, stipulate that "energy trading is only permitted with nodes whose power generation source is solar or wind energy".

[0167] Transaction conditions: Set economic and reputation thresholds for transactions. For example, set "the selling price shall not be lower than 0.5 yuan / kWh, and the purchase price shall not be higher than 1.0 yuan / kWh", and add "transactions are only allowed with nodes whose on-chain reputation score is higher than 85".

[0168] Step S2: Strategy Compilation and System Injection

[0169] User-defined advanced policies (configuration files or rules) are converted or compiled by the interface into standard instructions or parameter sets that the system kernel can recognize and execute. These compiled policy instructions are injected in real time into the decision logic of the following two core decision modules:

[0170] Inject personalized energy token pricing engines to influence their price calculations (such as the constraints of minimum selling price and maximum buying price in the pricing formula).

[0171] Inject a deterministic multi-hop energy balance controller to constrain its scheduling choices (e.g., the reputation threshold of the trading partner affects the path trust score, and the reserve capacity rule affects the available scheduling power).

[0172] Step S3: Policy Execution and Consistency Guarantee

[0173] In subsequent system operation, when the pricing engine performs transaction pricing or the energy balance controller performs path planning and scheduling, it will actively query and follow the injected strategy instructions corresponding to the user's assets. This ensures that every transaction and scheduling action automatically performed by the system always conforms to the pre-set economic benefit goals and risk preferences of the asset owner, achieving a unity between automated operation and personalized user control.

[0174] In summary, this application utilizes a proactive health inference module to perceive and assess the health status of each battery module in real time, and achieves unified modeling and coordinated scheduling of batteries with different chemical systems through a chemically neutral virtual battery abstraction layer. Based on this, a deterministic multi-hop energy balance controller predicts energy supply and demand and plans the optimal multi-hop transmission path based on a composite cost model, while a personalized energy token pricing engine dynamically generates transaction prices by integrating factors such as battery wear, system stability, and on-chain reputation, and completes automated settlement through blockchain smart contracts. Simultaneously, the system features dynamic security reconfiguration and user-programmable policy interfaces, supporting rapid fault isolation and the embedding of user-personalized policies. This solution achieves safe, economical, and reliable coordinated scheduling and decentralized trading of heterogeneous distributed energy storage systems, effectively improving asset lifespan, system resilience, and overall economic efficiency.

[0175] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent changes and alterations made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the scope of the present invention.

[0176] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention. Therefore, the embodiments should be regarded as exemplary and non-limiting in all respects.

[0177] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. An energy dispatch and decentralized transaction management system based on a battery energy storage system, characterized in that, include: A distributed battery energy storage system cluster consists of multiple physically dispersed, electrically isolated, and heterogeneous independent battery energy storage systems. Each battery energy storage system is equipped with a local controller, sensor array, and communication interface. The active health inference module is used to inject diagnostic pulses into the battery energy storage system and collect multimodal sensor data. It evaluates the health status of each battery module based on the Bayesian algorithm and uploads the results to the blockchain network. The chemically neutral virtual battery abstraction layer is used to establish chemical transfer functions for battery modules with different chemical systems, realize the mapping from power request to estimated efficiency, temperature rise and incremental wear, and perform chemically sensed virtual state of charge balancing and coordinated scheduling. A deterministic multi-hop energy balance controller is used to predict node energy deficits based on a causal state-space model, calculate composite transmission costs including power loss costs, battery wear costs, and trust penalties, and schedule the optimal multi-hop energy transmission path through a constrained optimization model. A personalized energy token pricing engine is used to dynamically generate personalized transaction prices based on battery wear and tear costs, system stability requirements, and on-chain reputation, and automatically execute settlement through blockchain smart contracts. The dynamic safety reconfiguration module is used to reconfigure the power topology through solid-state switches when a battery pre-runaway signal is detected, thereby achieving fault isolation and load transfer. The user-programmable policy interface is used to receive and execute user-defined charging and discharging policies, reserve capacity rules, energy sources, and trading conditions.

2. The energy dispatch and decentralized transaction management system based on a battery energy storage system according to claim 1, characterized in that, The health inference process performed by the active health detection module is as follows: periodically injecting diagnostic pulses with a duration of 1 to 5 seconds and an amplitude less than 10% of the rated current; simultaneously monitoring the voltage transient response, effective pulse impedance, battery surface and core temperature gradient, acoustic emission spectrum characteristics, micro-pressure or micro-strain drift, operating duty cycle, cumulative discharge depth, throughput, and HVAC system status during the pulse; and constructing a feature vector that includes the aforementioned voltage transient response, effective pulse impedance, battery surface and core temperature gradient, acoustic emission spectrum characteristics, micro-pressure or micro-strain drift, operating duty cycle, cumulative discharge depth, throughput, and HVAC system status. The posterior probability of each battery module being in a healthy, suspicious, or abnormal state is calculated based on Bayesian rules. The Bayesian rules are expressed as follows: ,in This is a prior probability based on historical data and the module's age. Let be the likelihood probability density of the observed eigenvectors under a given state. To observe the marginal probabilities of the feature vectors, a normalization constant is used; the posterior probabilities are normalized, and an evidence-weighted health index is calculated. , Preset weights for each state; Modules are classified according to the BEWHI index: A value ≥0.7 is considered a healthy state, and 0.4≤ A value <0.7 is considered suspicious. If the value is less than 0.4, it is considered an abnormal state, and the corresponding degraded operation, partial rebalancing, isolation, or maintenance work order instruction is triggered.

3. The energy dispatch and decentralized transaction management system based on a battery energy storage system according to claim 1, characterized in that, The cross-chemical scheduling optimization process implemented by the chemically neutral virtual battery abstraction layer is as follows: For battery modules with different chemical systems such as lithium iron phosphate, nickel manganese cobalt, sodium ion, and secondary batteries, chemical transfer functions are established respectively. The function maps the requested power Preq to a triplet ( , , ), representing the estimated efficiency, estimated temperature rise, and estimated incremental wear, respectively; The optimizer solves the objective function: Minimize:∑( + + ),in, To estimate the cost of electricity loss, Costs related to donor battery wear and tear. To mitigate the thermal risk penalty costs associated with the estimated temperature rise; the optimization process simultaneously satisfies the voltage window constraints, maximum charge / discharge current constraints, and overall system response time constraints of each module; by dynamically pairing battery modules with fast response characteristics with battery modules with high energy density but slower response, their power output is coordinated in the form of phase difference to suppress DC bus voltage deviation.

4. The energy dispatch and decentralized transaction management system based on a battery energy storage system according to claim 1, characterized in that, Composite unit energy transfer cost for path selection in deterministic multi-hop energy balance controllers Cost of power loss Cost of donor battery wear and trust penalties The sum of the three constitutes, that is ; Cost of power loss The path resistance loss and converter efficiency loss are monetized, and the calculation formula is as follows: ,in For reference energy prices, Energy is lost due to resistance. Energy loss due to inverter; cost of donor battery wear and tear. It reflects the marginal degradation caused by each kilowatt-hour of energy delivered, and its calculation is based on baseline throughput cost and operating point stress multiplier determined by temperature, charge / discharge rate, state of charge and depth of discharge; Trust Punishment The calculation depends on path trust score ,in Let h be the path hop count, γ be the hop count decay factor, d be the total path distance, and λ be the distance decay constant. The trust penalty function is: ,in, This is the preset maximum penalty amount.

5. The energy dispatch and decentralized transaction management system based on a battery energy storage system according to claim 1, characterized in that, Transaction prices generated by the personalized energy token pricing engine Settlement price from the underlying market Battery Wear Shadow Price (BWSP), Stability Reserve Premium (SRP), and Reputation Discount Joint decision, that is ; Battery wear shadow price (BWSP) is determined by the donor battery wear cost. Derived and superimposed with real-time thermal safety margin penalty; Stability Reserve Premium (SRP) is the pricing of the scarcity of the microgrid's current required rising reserve capacity, falling reserve capacity, and virtual inertia contribution to stability attributes, and its value originates from the Lagrangian dual variable or equivalent scarcity function of the system-level reserve constraint; Reputation Discount It is a function of the reliability score s on the donor node chain. ,and , This is the preset maximum discount limit.

6. The energy dispatch and decentralized transaction management system based on a battery energy storage system according to claim 1, characterized in that, Upon receiving a pre-runaway signal, the dynamic safety reconfiguration module executes a load redistribution strategy. This strategy calculates the optimal allocation scheme for the original load of the isolated module based on the proximity principle, the current load rate and health status index of each healthy module, and completes the dynamic reconfiguration of the power topology and seamless load switching within milliseconds by controlling the solid-state switch network.

7. The energy dispatch and decentralized transaction management system based on a battery energy storage system according to claim 1, characterized in that, The user-programmable policy interface receives asset owner policies including: charge / discharge policies, used to set charging or discharging during specific electricity price periods; reserve capacity rules, used to set the minimum reserve capacity percentage that must always be maintained; acceptable energy source rules, used to limit the energy type of the trading object; and transaction condition rules, used to set the minimum selling price, the maximum purchasing price, or the minimum credit score threshold of the trading object.

Images