CN-121998761-A - Virtual power plant transaction guarantee system and method with cooperative blockchain and brain-like calculation

CN121998761ACN 121998761 ACN121998761 ACN 121998761ACN-121998761-A

Abstract

The invention provides a virtual power plant transaction guarantee system and a virtual power plant transaction guarantee method with coordinated blockchain and brain-like computation, which creatively construct a four-layer progressive architecture, wherein a physical layer comprises a plurality of distributed energy nodes, each node is embedded with an edge proxy unit and is used for collecting real-time running state data, converting continuous analog quantity into discrete pulse signals to upload a nerve layer and receiving and executing upper-layer scheduling instructions, the nerve layer adopts an event-driven mechanism to carry out inter-node communication, triggers signal transmission when state variable changes exceed a threshold value, dynamically adjusts connection topology and weight according to node reputation and physical distance to realize edge intelligence and local coordination, a decision layer receives pulse characteristic aggregation information, respectively executes hierarchical scheduling computation in a plurality of layers to generate verified scheduling instructions to be issued step by step, and a protocol layer maintains a double-chain storage structure of an event chain and a value chain, verifies access identity, scheduling instructions and settlement data by utilizing intelligent contracts and combines an anti-quantum encryption protection data transmission link.

Inventors

GONG FEIXIANG
TIAN CHUANBO
CHENG FAN
LIU JING
SHI KUN
Luo dai
CHEN TIANTIAN
LI CHUNHONG
ZHANG ZIQIAN
SUN WENQIANG
XU XIAOYUAN

Assignees

中国电力科学研究院有限公司
国网上海市电力公司
东北大学
上海交通大学

Dates

Publication Date: 20260508
Application Date: 20251231

Claims (13)

1. A virtual power plant transaction assurance system with blockchain and brain-like computing coordination, comprising: The physical layer comprises a plurality of distributed energy nodes, an edge proxy unit is embedded in each distributed energy node, and the physical layer is used for collecting real-time running state data of equipment, converting the real-time running state data into discrete pulse signals, uploading the discrete pulse signals to the nerve layer, and receiving and executing scheduling instructions from an upper layer as a bottom layer sensing and executing end of the system; The neural layer is constructed on the physical layer based on a pulse neural network through the edge proxy unit, is used for carrying out inter-node communication by adopting an event driving mechanism, triggers signal transmission only when the state variable change exceeds a preset threshold, and dynamically adjusts connection topology and weight among nodes according to node reputation and physical distance; The decision layer is positioned on the upper layer of the nerve layer and connected with the nerve layer, the decision layer is used for receiving the feature convergence information uploaded by the nerve layer, respectively executing hierarchical scheduling calculation on a plurality of layers, generating verified scheduling instructions and gradually issuing the verified scheduling instructions to the physical layer through the nerve layer, the hierarchical scheduling calculation comprises edge instant optimization, regional collaborative optimization and central global optimization, and the feature convergence information is generated by performing feature extraction and summary processing on pulse signals through a pulse nerve network; And the protocol layer is positioned on the upper layer of the decision layer and is respectively connected with the nerve layer and the decision layer, and is used for maintaining a double-chain storage structure comprising an event chain and a value chain, verifying access identity, a scheduling instruction and settlement data by utilizing an intelligent contract, and encrypting and protecting data transmission links among all layers by combining an anti-quantum encryption technology.
2. The system of claim 1, wherein the access procedure of the node in the physical layer comprises: Initiating a registration request to a alliance chain certificate authority of a protocol layer, and receiving a digital certificate generated by the certificate authority, wherein the digital certificate is generated by registration information of a node; Generating a unique access credential based on the digital certificate and a hardware nerve fingerprint hash binding formed by combining node hardware features; Initiating an access request containing an anti-quantum signature to a protocol layer based on the unique access credential, and completing node access after passing the authentication of a protocol layer alliance chain authentication node And generating a unique access certificate through Ha Xibang based on the digital certificate and the hardware nerve fingerprint, and initiating an access request containing an anti-quantum signature to a protocol layer to realize anti-counterfeiting access of a physical node.
3. The system of claim 1, wherein the neural layer comprises a topology dynamic reconstruction module for updating connection weights between different nodes as calculated by: Wherein, the For a dynamic reputation score for node i, For the dynamic reputation score of node j, As the distance between the nodes of the network, In order for the attenuation coefficient to be a factor, Is the dynamic connection weight between node i and node j.
4. The system of claim 1, wherein the decision layer comprises: The edge real-time optimizing unit is embedded in the edge proxy unit and is used for processing the locally perceived pulse signals and the predicted data based on the event driving capability of the neural layer node, and generating control actions with the aim of minimizing economic cost and physical deviation; The regional collaborative optimization unit is used for exchanging key characteristic information through a pulse event network by utilizing the pulse characteristics of nodes in the aggregation region of the graph neural network, carrying out collaborative calculation between adjacent node clusters by adopting a distributed augmented Lagrange algorithm, and outputting a coordinated optimization decision; the central global optimization unit is used for integrating the regional optimization result and the power grid operation information, combining the full-grid constraint operation mixed type depth prediction and mathematical programming solver, solving the global optimization model, decomposing the optimization result into a strategy of issuing instructions and correcting the region and the edge.
5. The system of claim 4, wherein the calculation formula in the edge-on-demand optimization unit aimed at minimizing economic cost and physical deviation is as follows: Wherein, the Is a node Is used to adjust the decision vector of (a), Is the first Decision variables for individual nodes A corresponding economic cost function is provided for the system, Is the first The actual force exerted by the individual nodes is, Is issued by the upper layer The set-point values of the individual nodes, In order to penalize the coefficients, Is the first An objective function of the individual nodes; the calculation formula of the graph neural network aggregation in the region collaborative optimization unit is as follows: Wherein, the Is that Is used for the degree matrix of the (c), Is the function of the activation and, For the weight matrix of the first layer, In order to make an optimal decision after the coordination, As a feature of the first layer of the device, An augmented adjacency matrix that is the original adjacency matrix a; The central global optimization unit solves the global optimization model by the following calculation formula: Wherein, the In order to meet the requirements of the whole network, Is the first A set of physical constraints for the individual nodes, As a set of global decision variables, As a total cost of the system, To execute decision variables At the time of the first The cost of the individual distributed energy units, Is a feasible domain.
6. The system of claim 1, wherein the dual-chain storage structure of the protocol layer comprises: the event chain is used for storing the scheduling instruction hash and the abnormal alarm event generated by the high-frequency state pulse record and decision layer; a value chain for storing low frequency funds settlement records and reputation score changes; The event chain and the value chain are associated through encryption indexes, and a settlement mechanism is realized by combining intelligent contracts.
7. The system of claim 6, wherein the settlement mechanism is implemented by the following calculation formula: Wherein, the Is the first The value of the settlement of the individual nodes, Is the first The amount of the effective response of the individual nodes, As a reference price to be used, Is the first Personal node reputation Is used for the weight coefficient of the (c), Is based on the first Individual node response time Is used for the attenuation coefficient of the (c), Is the first Individual node execution accuracy Is used to determine the prize coefficient of (1), Is the first Special rewarding and punishing terms of the individual nodes.
8. The system of claim 1, wherein the protocol layer comprises an anti-quantum security module for: In the physical layer access and decision layer instruction issuing link, digital signature is carried out by using a digital signature algorithm based on a lattice; And in the link of establishing the data transmission link between all layers, establishing an encryption channel by using a grid-based key encapsulation mechanism, and adopting a distributed key generation technology to disperse and manage a system private key.
9. The system of claim 1, wherein the decision layer further comprises: the edge layer abnormality detection unit is used for calculating an error through the pulse self-encoder and judging that the node is abnormal when the error meets a preset abnormal condition; The regional layer immune response unit is used for calculating posterior probability after abnormality occurrence based on multi-node abnormality reports collected by the regional coordinator through Bayesian inference and merging evidence, and executing local immune response action after the calculated posterior probability exceeds a preset probability threshold; the central layer situation awareness unit is used for constructing a safety knowledge graph based on the on-chain records and the reputation index and identifying potential attack paths and hidden data injection behaviors.
10. The system of claim 9, wherein the error is calculated by the following equation: Wherein, the Is the original feature vector of the image and, Is the vector reconstructed from the encoder, Is an error; The posterior probability is calculated by the following calculation formula: Wherein, the For the posterior probability of the probability of a posterior, As a set of evidence that is to be used, For the purpose of the likelihood that the degree of the likelihood, For the a priori probabilities, Is the marginal probability of evidence.
11. A virtual power plant transaction guarantee method with cooperative blockchain and brain-like calculation is characterized by comprising the following steps: acquiring digital identity information and hardware physical characteristics of an application access node, generating a unique access credential through hash binding, and verifying the unique access credential based on a blockchain alliance chain to construct a trusted physical node set; Networking nodes in the trusted physical node set based on a pulse neural network algorithm, and dynamically reconstructing network topology according to node reputation and space-time attributes, wherein an event-driven mechanism is adopted for inter-node communication, and pulse signal generation and transmission are triggered only when state variable changes exceed a preset threshold value; Performing hierarchical scheduling calculation according to the characteristic information carried by the pulse signals, generating verified scheduling instructions and issuing the verified scheduling instructions to be performed, wherein the hierarchical scheduling calculation comprises scheduling calculation from immediate response of an edge side, collaborative optimization of an area side to global overall planning of a center side; and classifying and storing event data and value data in the dispatching calculation process by using a double-chain architecture, executing multidimensional value settlement based on intelligent contracts, and encrypting the whole flow data by combining an anti-quantum encryption technology.
12. The electronic equipment is characterized by comprising at least one processor and a memory, wherein the memory and the processor are connected through a bus; The memory is used for storing one or more programs; The virtual power plant transaction assurance method of blockchain and brain-like computing cooperation of claim 11, when the one or more programs are executed by the at least one processor.
13. A readable storage medium having stored thereon an execution program which, when executed, implements the virtual power plant transaction assurance method of blockchain and brain-like computing coordination of claim 11.

Description

Virtual power plant transaction guarantee system and method with cooperative blockchain and brain-like calculation Technical Field The invention relates to the field of computers, in particular to a virtual power plant transaction guarantee system and method with cooperative blockchain and brain-like calculation. Background In recent years, global energy systems are undergoing profound structural transformations. With the rapid development of renewable energy sources, distributed power generation and energy storage technologies, the operation mode of an energy system has the characteristics of high distribution, dynamics and intellectualization. The virtual power plant is used as a novel energy management and scheduling mode, and distributed energy resources of different types and different types are aggregated into a whole which can be scheduled and optimized through advanced communication, control and information processing technologies so as to realize unified coordination and efficient operation. This model has shown great potential in improving renewable energy utilization, participating in market transactions and auxiliary services, and has gradually become an important component of the energy internet. However, the existing virtual power plant system still has a plurality of systematic bottlenecks under the background of large-scale new energy access and multi-party cooperative operation. Firstly, in terms of data credibility, a traditional virtual power plant generally relies on a centralized data acquisition and reporting mechanism, and a distributed energy terminal needs to upload operation data to a central server for unified processing. In the mode, the authenticity of the data completely depends on the credibility of the central node, a tamper-proof and verification mechanism crossing the participants is lacking, and the phenomena of node false report, delayed report or fake running state are easy to occur. Data distortion not only affects the accuracy of scheduling and control, but may also raise trust crisis and transaction disputes in the multi-subject market. Secondly, in terms of real-time performance, the fluctuation of the new energy output requires that the scheduling system has response capacity of seconds or even milliseconds. However, with the expansion of the node scale, the computation complexity of the centralized optimization model increases sharply, and the central server often needs several minutes or even longer to generate a scheduling scheme, which cannot meet the dynamic response requirement in the high-permeability new energy scene. Meanwhile, the time delay of the common block chain consensus mechanism is difficult to match with real-time regulation and control, so that the scheduling delay of the whole system is difficult to be reduced to the engineering usable range. In addition, conventional virtual power plant architectures typically operate in a "central convergence-unified computing" mode in terms of synergistic efficiency and scalability. When the number of the nodes reaches tens of thousands, the overall optimization calculation amount is increased in a nonlinear manner, the load of data transmission and network communication is increased sharply, and communication congestion and scheduling lag are easy to cause. Due to the lack of hierarchical optimization and edge autonomous mechanisms, local rapid optimization and global coordination and unification are difficult to achieve among different geographic areas and different types of distributed resources, and the flexibility and response speed of the system are obviously limited. In the aspect of safety protection, the geographic distribution of the distributed energy nodes is wide and various, and the distributed energy nodes are very easy to be targets of network attacks. Once subjected to data injection, glitches or denial of service attacks, local or even global regulatory failures may be triggered. The safety protection means of the existing system is still mainly based on the traditional public key encryption system, and a coping mechanism is lacking for safety threat brought by future quantum computation. Meanwhile, the system lacks a layered active defense and cooperative immune mechanism, once a single point is damaged, a linkage effect is easily caused, and potential safety hazards of the system level are caused. More importantly, the existing research is concentrated on improvement of algorithm level or local functional modules, such as economic dispatch, load prediction or aggregate transaction model, etc., but a unified system architecture framework is not formed yet, and a top-level design from the system engineering level is lacking. The traditional system is often advanced in algorithm and lagged in architecture, and cannot realize cross-level collaborative optimization and full-link trusted operation when facing to an energy internet scene of complex environment and multi-party interaction, and is