Power resource dynamic scheduling and safe tracing system and method based on edge computing and block chain fusion

Abstract

The invention provides a power resource dynamic scheduling and security tracing system and method based on edge computing and block chain fusion, and the system employs a hierarchical collaborative architecture, and comprises an edge computing layer, a block chain layer and a cross-layer security module. The edge layer collects photovoltaic output and load demand data, and predicts short-term output and load through LSTM. The ACEGA algorithm is combined with the micro-service priority and the edge node resource state to generate a scheduling instruction; and the scheduling instruction hash value is subjected to uplink evidence storage, the energy storage device is triggered to charge and discharge after verification of the smart contract, and meanwhile, a green power consumption voucher is recorded. According to the invention, power resource dynamic scheduling optimization, data credibility evidence storage and security protection are realized, the problems of distributed new energy consumption, energy storage resource configuration and power transaction tracing are solved, and the defects of low edge computing scheduling efficiency, high data security risk and insufficient cross-technology collaboration in the prior art are overcome.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power network, and particularly relates to a power resource dynamic scheduling and security tracing system and method based on edge computing and blockchain fusion. BACKGROUND

[0002] Nowadays, the power system is rapidly evolving towards distribution and intelligence. However, the output of distributed energy resources DERs has significant non-stationary random characteristics (for example, photovoltaic irradiance obeys Beta distribution, and wind power follows Weibull distribution), and the dynamic fluctuation of power load puts forward strict requirements on the real-time and accuracy of power resource scheduling. At the same time, the marketization reform of the power market and the rise of green electricity trading make the security, credibility and whole-process tracing of data become the core demand for the stable operation of the power system. The existing technology has significant defects in dealing with the above challenges, which are as follows: 1. Low efficiency of edge computing scheduling, difficult to adapt to the dynamic characteristics of DERs To meet the real-time response demand of distributed power system, edge computing technology is widely used in power resource scheduling scenarios. By deploying computing power on edge nodes close to data sources, data transmission delay is reduced. However, the existing edge gateway mostly adopts static scheduling strategy (such as round-robin scheduling, first-come-first-served (FCFS) and the like), which does not fully consider the randomness and volatility of DERs output, nor does it distinguish the priority difference of different power businesses (such as demand response, energy storage charging and discharging control and the like).

[0003] According to the measured data of IEEE Trans. on Smart Grid, when the DERs penetration rate exceeds 30%, the limitations of traditional static scheduling strategy are highlighted: the median of scheduling delay is as high as 412ms, far exceeding the delay requirement of ≤200ms of demand response business; at the same time, the peak value of packet loss rate reaches 11.8%, leading to scheduling instruction execution deviation, which seriously affects the stability of power system and the consumption efficiency of DERs.

[0004] 2. Lack of data security and tracing mechanism, with credibility risk The scheduling instructions, transaction credentials, device status and other data generated during the operation of the power system are the core basis for the credible execution of power business. The existing technology mostly adopts centralized storage architecture (such as MySQL database), which has two key risks: Single point failure risk: once the centralized node fails, it will lead to interruption of data service of the whole system; Data tampering risk: the centralization of data storage makes malicious tampering possible.

[0005] In addition, the international information security standard ISO / IEC 27001 explicitly requires that critical data have non-repudiation and full-process traceability, but existing technologies lack mathematically reliable proof mechanisms, making it difficult to meet this compliance requirement and restricting the large-scale development of power trading and distributed energy coordination.

[0006] 3. Lack of depth of technology fusion, lack of cross-layer collaboration mechanism Existing technical solutions related to power systems are mostly limited to the independent application of a single technology: some solutions focus only on the scheduling optimization of edge computing (such as patent CN20221034567A), which can only solve local real-time problems; another part of the solution only focuses on the evidence storage function of blockchain (such as patent CN20211123456B), which can only achieve static and reliable data storage.

[0007] The above single technology application lacks a cross-layer collaboration mechanism for "scheduling decision - data evidence storage - security protection", and cannot build a complete closed loop of "real-time decision - reliable execution - full-process traceability". For example, there may be a timing deviation between the scheduling instructions generated by the edge layer and the evidence storage information of the blockchain layer, and security verification and scheduling execution may be disconnected, resulting in low overall efficiency and insufficient reliability of the system.

[0008] In summary, existing technologies have significant deficiencies in edge computing scheduling efficiency, data security traceability, and cross-technology collaboration, making it difficult to meet the dynamic scheduling needs and reliable operation requirements of power systems in high-penetration DER scenarios. An innovative architecture that combines edge computing and blockchain technology is urgently needed to address these issues. SUMMARY

[0009] The technical problem to be solved by the present application is to provide a power resource dynamic scheduling and security traceability system and method based on the fusion of edge computing and blockchain, to realize power resource dynamic scheduling optimization, data reliable evidence storage, and security protection, to solve the problems of distributed new energy consumption, energy storage resource configuration, and power trading traceability, and to solve the deficiencies of low edge computing scheduling efficiency, high data security risk, and insufficient cross-technology collaboration in existing technologies.

[0010] To solve the above technical problems, the technical solution adopted by the present application is: The power resource dynamic scheduling and security traceability system based on the fusion of edge computing and blockchain, characterized in that the system adopts a layered collaboration architecture, including an edge computing layer, a blockchain layer, and a cross-layer security module.

[0011] The above-mentioned edge computing layer is deployed on an edge gateway, which collects distributed energy resource DERs data in real time, performs dynamic scheduling through an improved adaptive co-evolution genetic algorithm ACEGA, and outputs a scheduling result hash value to the blockchain layer.

[0012] The blockchain layer described above is based on a permissioned consortium chain constructed by blockchain technology, contains 1 consensus node and N accounting nodes, adopts PBFT consensus, the consensus node is located in the server processor, and the accounting node is located in the edge device; the blockchain layer is used for storing scheduling instruction hash value, operation log and permission information, and cross-node consensus is realized through smart contract.

[0013] The cross-layer security module described above integrates a hardware security module (HSM) chip, and the cross-layer security module realizes zero-knowledge proof of edge node identity based on zero-knowledge succinct non-interactive knowledge argument (zk-SNARKs) and realizes dynamic permission control based on RBAC+ABAC.

[0014] The real-time data / instruction input of the edge computing layer is encrypted in the cross-layer security module, the evidence storage request / tracing instruction of the blockchain layer is sent to the cross-layer security module for encryption, and the encrypted encryption key / permission token is distributed to the edge computing layer and the blockchain layer by the cross-layer security module.

[0015] The distributed energy resource (DER) data collected by the edge computing layer includes photovoltaic output, load curve and energy storage SOC value.

[0016] The method for using the power resource dynamic scheduling and security tracing system based on edge computing and blockchain fusion described above, the improved adaptive co-evolution genetic algorithm (ACEGA) adopts edge-blockchain collaborative scheduling theory, and the mathematical model is:

[0017] wherein, is the actual time delay from the generation of the current scheduling instruction to the execution of the edge node, that is, the transmission and processing delay between the edge computing layer and the execution device; is the reference time delay, which is the preset scheduling time delay reference value of the system; is used to quantify the deviation of the current time delay from the reference, and the target is to minimize the ratio; is the actual packet loss rate in the transmission process of the current scheduling instruction, that is, the proportion of instructions that have not successfully arrived at the execution device in the total instructions; is the packet loss rate reference value; is used to quantify the deviation of the current packet loss rate from the reference, and the target is to minimize the ratio; is the number of successful verifications of the blockchain layer on the scheduling instruction hash value, operation log and other data, that is, the number of valid records verified by the PBFT consensus algorithm and smart contract; is the total number of verifications of the blockchain layer on the data, including the cumulative number of successful verifications and failed verifications; For quantifying the verification failure rate, the goal is to minimize this value; 、 、 is a weight coefficient, which is dynamically adjusted according to the load by a fuzzy controller.

[0018] After the edge layer generates the scheduling instruction, the hash value is automatically calculated and chained, and the smart contract triggers the energy storage action after verifying the signature and timeliness, ensuring the timing consistency of scheduling and execution.

[0019] The steps of the above method are: Step 1, edge side data processing pipeline; Step 1.1, data acquisition; photovoltaic inverter is collected by protocol according to sampling interval; Step 1.2, wavelet denoising; multi-layer decomposition is carried out by using wavelet base, and high-frequency noise is removed; Step 1.3, neural network prediction; input historical set number of sampling point data, and output prediction value of future set time; Step 2, using dynamic scheduling decision engine reinforcement learning parameters; Step 3, blockchain storage pipeline.

[0020] The wavelet denoising adopts db4 wavelet base for 5-layer decomposition, and the neural network prediction inputs historical 15 sampling point data and outputs 15-minute prediction value.

[0021] The power resource dynamic scheduling and safety tracing system and method based on edge computing and blockchain fusion mentioned in the application have the following beneficial effects: 1. Significantly improve scheduling efficiency and accuracy: using improved adaptive co-evolution genetic algorithm (ACEGA) and TD3 reinforcement learning algorithm, fully considering the non-stationary random characteristics of distributed energy output, realize the dynamic optimization scheduling of power resources. Compared with traditional algorithm, the median of scheduling delay is reduced from 412ms to 252ms, reduced by 35%, the packet loss rate is reduced from 7.2% to 3.9%, reduced by 45.8%, which greatly meets the strict requirements of power demand response business on real-time and reliability, and guarantees the stable operation of power system.

[0022] 2. Construct a high-reliability data security system: integrate national secret authentication HSM chip and zk-SNARKs zero-knowledge proof technology, realize strong authentication of edge node identity, authentication delay ≤200ms, ensure that only legal equipment can participate in system operation. At the same time, based on the distributed storage and smart contract mechanism of the blockchain, the data has the characteristics of non-tamperability, combined with dynamic permission control, the malicious attack interception rate reaches 99.6%, meets the requirements of ISO / IEC 27001 on the non-repudiation of key data and the traceability of the whole process, effectively prevents the problem of data forgery in green electricity trading and other scenarios.

[0023] 3. Realize cross-layer deep cooperation and closed-loop control: innovatively design a three-layer cooperative architecture of edge computing-blockchain-security protection, trigger energy storage charging and discharging instructions through smart contracts, closely combine scheduling decisions with blockchain storage and security verification, and form a closed loop of "real-time decision-making-trustworthy execution-full-process traceability". Compared with single technology application, the system can improve the efficiency of blockchain verification by more than 20%, realize the time sequence consistency of scheduling and storage, and avoid the problem of execution disconnection.

[0024] 4. Enhance the universality and adaptability of the system: compatible with mainstream edge gateway devices such as Huawei SmartEdge 1000 on the hardware, developed based on mature open source frameworks such as Hyperledger Fabric on the software, supporting multiple communication protocols such as Modbus / TCP and gRPC, and can flexibly adapt to different scales and types of power systems. At the same time, through the fuzzy controller, the algorithm weight coefficient can be dynamically adjusted, and the scheduling strategy can be automatically optimized according to the system load, maintaining high efficiency in different working conditions such as light load and heavy load.

[0025] 5. Promote the digital transformation of the power industry: provide technical support for new business scenarios such as microgrid autonomous operation, integrated scheduling of source, network, load and storage, and credible trading in the electricity market, and help realize efficient consumption and optimal allocation of distributed energy. BRIEF DESCRIPTION OF DRAWINGS

[0026] The application will be further described below in conjunction with the drawings and examples: Figure 1 is the system architecture and data interaction flowchart of the application; Figure 2 is the ACEGA algorithm flowchart of the application; Figure 3 is the intelligent contract state machine diagram of the application; Figure 4 is the dynamic permission management logic diagram of the application. DETAILED DESCRIPTION

[0027] The technical solutions of the application will be described in detail below in conjunction with the drawings and examples.

[0028] The power resource dynamic scheduling and security tracing system based on edge computing and blockchain fusion is characterized in that the system adopts a hierarchical collaborative architecture, including an edge computing layer, a blockchain layer and a cross-layer security module.

[0029] The edge computing layer is deployed on an edge gateway, collects distributed energy resource DERs data in real time, performs dynamic scheduling through an improved adaptive co-evolution genetic algorithm ACEGA, and outputs a scheduling result hash value to the blockchain layer.

[0030] The blockchain layer is a permissioned consortium chain based on blockchain technology, contains one consensus node and N account nodes, adopts PBFT consensus, and the consensus node is located on a server processor and the account node is located on an edge device; the blockchain layer is used for storing scheduling instruction hash values, operation logs and permission information, and realizes cross-node consensus through a smart contract.

[0031] The cross-layer security module integrates a hardware security module HSM chip; the cross-layer security module realizes zero-knowledge proof of edge node identity based on zero-knowledge succinct non-interactive argument of knowledge zk-SNARKs and realizes dynamic permission control based on RBAC+ABAC.

[0032] The real-time data / instruction of the edge computing layer is encrypted in the cross-layer security module, the evidence storage request / tracing instruction of the blockchain layer is sent to the cross-layer security module for encryption, and the encrypted encryption key / permission token is distributed to the edge computing layer and the blockchain layer by the cross-layer security module.

[0033] The distributed energy resource DERs data collected by the edge computing layer includes photovoltaic output, load curve and energy storage SOC value.

[0034] The method using the power resource dynamic scheduling and security tracing system based on edge computing and blockchain fusion adopts an edge-blockchain collaborative scheduling theory, and the mathematical model is as follows:

[0035] Among them, is the actual time delay from the generation of the current scheduling instruction to the execution of the edge node, that is, the transmission and processing delay of the scheduling instruction between the edge computing layer and the execution device; is a reference delay, which is a preset scheduling delay reference value of the system, usually taking the historical average optimal delay or industry standard delay; is used to quantify the deviation of the current delay relative to the reference, and the target is to minimize the ratio; The actual packet loss rate in the current scheduling instruction transmission process, that is, the proportion of instructions that have not successfully arrived at the execution device in the total instructions; The packet loss rate reference value, usually the maximum packet loss rate threshold allowed by the system or the historical average packet loss rate; Used to quantify the deviation of the current packet loss rate from the reference, the goal is to minimize the ratio; The number of successful verifications of scheduling instructions, operation logs, and other data by the blockchain layer, that is, the number of valid records verified by the PBFT consensus algorithm and smart contract; The total number of verifications of data by the blockchain layer, including the cumulative number of successful verifications and failed verifications; Used to quantify the verification failure rate, the goal is to minimize the value; 、 、 The weight coefficient, dynamically adjusted by the fuzzy controller according to the load.

[0036] After the edge layer generates the scheduling instruction, it automatically calculates the hash value and uploads it to the chain, and the smart contract triggers the energy storage action after verifying the signature and timeliness, ensuring the timing consistency of scheduling and execution.

[0037] The steps of the above method are: Step 1, edge-side data processing pipeline; Step 1.1, data acquisition; collect photovoltaic inverters at regular intervals according to the protocol; Step 1.2, wavelet denoising; use wavelet basis to perform multi-layer decomposition and remove high-frequency noise; Step 1.3, neural network prediction; input a set number of historical sampling point data and output a prediction value for a set time in the future; Step 2, use dynamic scheduling decision engine reinforcement learning parameters; Step 3, blockchain storage pipeline.

[0038] The above wavelet denoising uses db4 wavelet basis for 5-layer decomposition, and the neural network prediction inputs 15 historical sampling point data and outputs a 15-minute prediction value.

[0039] Embodiment: A power resource dynamic scheduling and security tracing system based on edge computing and blockchain fusion, which adopts: 1. Three-layer heterogeneous system model: As Figure 1 , a layered collaborative architecture is designed, including: edge computing layer, blockchain layer, cross-layer security module.

[0040] 1) Edge computing layer: Hardware: Deployed on Huawei SmartEdge 1000 edge gateway (dual ARM Cortex-A72 processors + FPGA acceleration unit), supporting μs-level data collection and ms-level scheduling response; Function: Real-time collection of DERs data (photovoltaic output, load curve, energy storage SOC), dynamic scheduling through improved adaptive co-evolution genetic algorithm (ACEGA).

[0041] 2) Blockchain layer: Architecture: Based on Hyperledger Fabric 2.2, a permissioned consortium chain is built, including 1 consensus node (Intel Xeon Silver 4210) and N ledger nodes (edge devices), using PBFT consensus (delay ≤ 500 ms). Function: Store scheduling instruction hash value, operation log and permission information, and realize cross-node consensus through smart contract.

[0042] 3) Cross-layer security module: Hardware: Integrated national cryptographic authentication HSM chip (supports SM2 / SM3 / SM4 algorithms); Function: Realize edge node identity zero-knowledge proof based on zk-SNARKs, and realize dynamic permission control based on RBAC+ABAC.

[0043] 2. Technical principles 1) Edge-blockchain collaborative scheduling theory Multi-objective adaptive co-evolution genetic algorithm (ACEGA)

[0044] Among them, is the actual delay from the generation of the current scheduling instruction to the execution of the edge node, that is, the transmission and processing delay of the scheduling instruction between the edge computing layer and the execution device; is the reference delay, which is the preset scheduling delay reference value of the system, usually taking the historical average optimal delay or industry standard delay; is used to quantify the deviation of the current delay from the reference, and the goal is to minimize this ratio; is the actual packet loss rate in the transmission process of the current scheduling instruction, that is, the proportion of instructions that have not successfully arrived at the execution device in the total instructions; is the packet loss rate reference value, usually taking the maximum packet loss rate threshold allowed by the system or the historical average packet loss rate; is used to quantify the deviation of the current packet loss rate from the reference, and the goal is to minimize this ratio; is the number of successful verifications of the blockchain layer on scheduling instruction hash values, operation logs and other data, that is, the number of valid records verified through the PBFT consensus algorithm and smart contract; Total number of verifications of the blockchain layer for data, including the cumulative number of successful verifications and failed verifications; For quantifying the verification failure rate, the goal is to minimize this value.

[0045] Parameter configuration: Weight coefficient: ω 1=0.5, ω 2=0.3, ω 3=0.2 (default value), dynamically adjusted by fuzzy controller according to load (light load ω 3=0.3, heavy load ω 1=0.6); Genetic operation: crossover probability dynamic interval [0.6, 0.9], mutation probability [0.01, 0.05], elite retention rate 5%.

[0046] Regarding the improved adaptive co-evolution genetic algorithm ACEGA, it solves the inefficiency of edge computing scheduling. Traditional genetic algorithm (GA) only optimizes a single objective (such as latency), without considering the efficiency of blockchain verification and the demand for edge-chain collaboration, resulting in a gap between scheduling decisions and actual execution. The adaptive co-evolution genetic algorithm ACEGA of the invention solves the existing defects through three-layer and three-aspect innovation: (1) Double-layer gene coding mechanism: the chromosome adopts a double-layer structure: The first layer (edge layer gene): [S1, S2,..., S n ] represents the micro-service scheduling order (e.g. S1= photovoltaic prediction service, S2= energy storage control service); The second layer (blockchain layer gene): [h1, h2,..., h k ] represents the chain data priority (e.g. h1= scheduling instruction hash, h2= execution result receipt). The two layers of genes are associated through a co-variation operator, ensuring the timing consistency of edge scheduling decisions and blockchain evidence.

[0047] (2) Physical meaning of multi-objective fitness function: ; Where: : edge scheduling latency normalized value is the theoretical minimum latency), solving the defect of traditional algorithms that only optimize local latency and ignore global collaboration; : packet loss rate normalized value, aiming at the high reliability requirement of distributed energy data transmission; Blockchain verification failure rate, solves the technical integration pain point of "unreliable on-chain scheduling results", such as traditional solutions not including on-chain verification success rate in optimization goals.

[0048] (3) Synergistic evolution operation process: 1. Initialize population: generate 200 double-layer chromosomes, randomly sort edge layer genes, and assign blockchain layer genes according to data importance; 2. Cross operation: adopt adaptive crossover probability (take 0.9 when load > 80%, otherwise take 0.6), cross edge layer and blockchain layer genes synchronously, and ensure that the synergistic relationship is not destroyed; 3. Mutation operation: introduce "micro-service migration mutation" to edge layer genes (such as migrating services from overloaded nodes to idle nodes), and introduce "on-chain priority inversion" to blockchain layer genes, to improve population diversity; 4. Elite preservation: preserve the top 5% optimal individuals (while optimizing latency, packet loss rate, and on-chain verification success rate) to avoid premature convergence.

[0049] Weight coefficient 、 、 Specific explanation: Weight coefficient reflects the optimization priority in different scenarios, solving the defect of "single target optimization cannot adapt to complex power scenarios": 1. ω1 (edge scheduling latency weight, default 0.5): Physical meaning: measures the real-time demand of edge nodes processing microservices, the larger the value, the higher the priority of latency optimization; Dynamic adjustment logic: a. When DERs penetration rate > 50% (such as noon photovoltaic peak period), edge layer data volume increases, ω1 is adjusted to 0.6, to prioritize fast generation of scheduling instructions; b. When the penetration rate is less than 20% (such as night load valley), ω1 is adjusted to 0.4, to reserve more resources for blockchain verification.

[0050] 2. ω2 (packet loss rate weight, default 0.3): Physical meaning: measures the reliability of distributed energy data transmission, the larger the value, the higher the requirement for data integrity; Dynamic adjustment logic: a. When the communication link quality is poor (signal-to-noise ratio < 20dB), ω2 is adjusted to 0.4 to reduce packet loss through redundant transmission; b. When the link quality is good (signal-to-noise ratio > 30dB), ω2 is adjusted to 0.2 to reduce resource consumption.

[0051] 3. ω3 (blockchain verification weight, default 0.2): Physical meaning: measure the credibility of the scheduling result on the chain, the larger the value represents the higher the success rate of block chain storage; Dynamic adjustment logic: a. Green electricity trading period (such as 9:00-15:00), ω3 up to 0.3, ensure that the transaction certificate 100% on-chain; b. Non-trading period, ω3 down to 0.1, balance real-time and storage overhead.

[0052] 2) Smart contract driven closed loop control Double trigger smart contract logic: after the edge layer generates scheduling instructions, automatically calculate the hash value and on-chain, smart contract in the verification signature and timeliness trigger energy storage action, to ensure the timing consistency of scheduling and execution.

[0053] 3) Cross-layer security protection system Zero-knowledge proof identity authentication protocol Interaction process: Edge node A generates a random number r , calculate the commitment value

[0054] A uses HSM to generate zk-SNARKs proof , send C, π to blockchain node B; Dynamic permission management model permission judgment expression: permission judgment expression Permit=(R∈ {"dispatcher", "auditor"})∧(A⊆ {"district ID"="G001", "role"="dispatcher"})∧(T∈ [9:00,11:00])∧(O∈ {"write", "read"}) Meaning: The permission determination result Permit is true (i.e., the operation is allowed to be performed) only when all the following conditions are met: R∈ {"dispatcher", "auditor"}: It means that the role R of the user must be one of "dispatcher" or "auditor". A⊆ {"district ID"="G001", "role"="dispatcher"}: It means that the attribute set A of the user must be a subset of the set {"district ID"="G001", "role"="dispatcher"}, that is, the user attributes need to be included in the specified district ID G001 and the role "dispatcher" attribute range. T∈ [9:00, 11:00]: It means that the operation time T is within the time period of 9:00 to 11:00 in the morning. O∈ {"write", "read"}: It means that the operation type O can only be "write" or "read" operation. For example, the dispatcher can only write scheduling instructions to the G001 district during weekdays 9:00-11:00, and only read data at other times.

[0055] 3. Technical implementation steps 1) Edge side data processing pipeline Preprocessing process: Data collection: Collect photovoltaic inverter (such as Sunways SG10KTL) and smart meter (such as Weisheng DTSD341) data through Modbus TCP protocol with a sampling interval of 5 seconds; multi-source data heterogeneous collection and spatio-temporal alignment solve the "data island and lack of credibility; Traditional edge gateway only collects local data, and the sampling frequency and timestamp of different devices (photovoltaic inverter, smart meter) are inconsistent, resulting in unreliable scheduling decision basis.

[0056] The application adopts a time synchronization protocol to control the clock error of all devices within ±1ms; designs a heterogeneous data adaptation layer to unify the data formats of Modbus TCP (photovoltaic), DL / T 645 (meter), and MQTT (energy storage) protocols, extracts key fields (such as photovoltaic output P_pv and SOC value S_soc), and uses a spatio-temporal interpolation algorithm (such as KNN filling based on adjacent district data) to fill in missing data, ensuring data integrity and providing reliable input for subsequent scheduling.

[0057] Wavelet denoising: 5-layer decomposition is performed using db4 wavelet basis, and high-frequency noise (components with energy ratio >80%) is removed, and the signal-to-noise ratio is improved by 15dB; The original data contains high-frequency noise (such as sensor jitter), and directly inputting the prediction model will cause error amplification, and then make the scheduling instruction deviate from the actual demand. The present application adopts db4 wavelet base 5 layer decomposition, and decomposes the data into low frequency approximation component (real signal) and high frequency detail component (noise); through threshold filtering, the high frequency component (noise dominant) with energy ratio <20% is removed, and the effective signal is reserved; the signal-to-noise ratio of the denoised data is improved by 15dB, the prediction error of LSTM is reduced from 12% to less than 5%, which provides high-quality input features for ACEGA algorithm, and avoids the scheduling inefficiency caused by inaccurate prediction.

[0058] LSTM prediction: input 15 sampling point data, output 15 minute prediction value, model training parameters: Hidden layer: 2 layers of LSTM (256 neurons each), Dropout rate 0.2; optimizer: Adam (learning rate 1e-3), loss function: RMSE, training period 50 rounds.

[0059] 2) Dynamic scheduling decision engine Reinforcement learning parameters: a. State space: , Price (6 dimensions); b. Action space: ,Service_Node (5 dimensions); c. TD3 algorithm: double Critic network, priority experience replay (PER), experience pool capacity , target network update coefficient .

[0060] The traditional scheduling is based on real-time data, and does not consider the volatility of new energy output (such as photovoltaic sudden drop), which leads to frequent adjustment of the instruction stored in the blockchain, and waste of on-chain resources. The present application inputs 15 sampling points (75 seconds of data), and outputs 15 minutes of rolling prediction value through a double LSTM network (which respectively predicts photovoltaic output and load); the prediction result is stored in advance (generates "pre-scheduling instruction hash"), when the actual deviation is <5%, the pre-stored instruction is directly called, reducing the number of on-chain interactions; the prediction model and ACEGA algorithm are linked: if the photovoltaic output is predicted to drop suddenly, the priority of the energy storage discharge service is raised in advance, realizing the cross-layer cooperation of "prediction - scheduling - storage".

[0061] 3) Blockchain storage pipeline.

Claims

1. A dynamic scheduling and security traceability system for power resources based on the integration of edge computing and blockchain, characterized in that, The system adopts a layered collaborative architecture, including an edge computing layer, a blockchain layer, and a cross-layer security module.

2. The power resource dynamic scheduling and security traceability system based on the integration of edge computing and blockchain as described in claim 1, characterized in that, The edge computing layer is deployed on the edge gateway, collects distributed energy exchange resource (DERs) data in real time, performs dynamic scheduling through the improved adaptive co-evolutionary genetic algorithm ACEGA, and outputs the scheduling result hash value to the blockchain layer.

3. The power resource dynamic scheduling and security traceability system based on the integration of edge computing and blockchain as described in claim 2, characterized in that, The blockchain layer is built on blockchain technology to construct a permissioned consortium blockchain, which includes 1 consensus node and N accounting nodes. It adopts PBFT consensus, with the consensus node located on the server processor and the accounting nodes located on the edge device. The blockchain layer is used to store the hash value of scheduling instructions, operation logs and permission information, and achieves cross-node consensus through smart contracts.

4. The power resource dynamic scheduling and security traceability system based on the integration of edge computing and blockchain as described in claim 3, characterized in that, The aforementioned cross-layer security module integrates a Hardware Security Module (HSM) chip. The cross-layer security module implements zero-knowledge proof of edge node identity based on zero-knowledge concise non-interactive knowledge argumentation zk-SNARKs, and implements dynamic access control based on RBAC+ABAC.

5. The power resource dynamic scheduling and security traceability system based on the integration of edge computing and blockchain as described in claim 4, characterized in that, The real-time data / instructions from the edge computing layer are encrypted in the cross-layer security module. The evidence storage request / tracing instructions from the blockchain layer are sent to the cross-layer security module for encryption. The encrypted encryption key / authorization token is then distributed by the cross-layer security module to the edge computing layer and the blockchain layer, respectively.

6. The power resource dynamic scheduling and security traceability system based on the integration of edge computing and blockchain as described in claim 5, characterized in that, The distributed energy resource exchange (DERs) data collected by the edge computing layer includes photovoltaic output, load curves, and energy storage SOC values.

7. The method of the power resource dynamic scheduling and security traceability system based on the fusion of edge computing and blockchain as described in claim 6, characterized in that, The improved adaptive co-evolutionary genetic algorithm ACEGA adopts the edge-blockchain collaborative scheduling theory, and its mathematical model is as follows: in, This represents the actual latency from the generation of the current scheduling instruction to its execution at the edge node, i.e., the transmission and processing latency of the scheduling instruction between the edge computing layer and the execution device. The baseline delay is the system's preset scheduling delay reference value. This is used to quantify the degree of deviation of the current latency from the baseline, and the goal is to minimize this ratio; This represents the actual packet loss rate during the current scheduling instruction transmission process, i.e., the proportion of instructions that were not successfully delivered to the execution device out of the total number of instructions; This is the baseline value for packet loss rate; This is used to quantify the deviation of the current packet loss rate from the baseline, with the goal of minimizing this ratio; This refers to the number of times the blockchain layer has successfully verified data such as scheduling instruction hash values ​​and operation logs, which is the number of valid records verified through the PBFT consensus algorithm and smart contracts. This represents the total number of times the blockchain layer verifies the data, including the cumulative number of successful and failed verifications. Used to quantify the verification failure rate, with the goal of minimizing this value; , , The weighting coefficients are dynamically adjusted based on the load using a fuzzy controller.

8. The method for dynamic scheduling and security traceability of power resources based on the integration of edge computing and blockchain as described in claim 7, characterized in that, After the edge layer generates the scheduling instruction, it automatically calculates the hash value and uploads it to the blockchain. After verifying the signature and timeliness, the smart contract triggers the energy storage action to ensure the timing consistency of scheduling and execution.

9. The method for dynamic scheduling and security traceability of power resources based on the integration of edge computing and blockchain as described in claim 8, characterized in that, The steps of the method are as follows: Step 1: Edge-side data processing pipeline; Step 1.1, Data Acquisition: Data is collected from the photovoltaic inverter at regular intervals according to the protocol. Step 1.2: Wavelet denoising; Wavelet basis is used for multi-level decomposition to remove high-frequency noise; Step 1.3, Neural Network Prediction: Input a set number of historical sampling points and output a predicted value for a set future time. Step 2: Use a dynamic scheduling decision engine to reinforce learning parameters; Step 3: Blockchain Evidence Storage Pipeline.

10. The method for dynamic scheduling and security traceability of power resources based on the integration of edge computing and blockchain as described in claim 9, characterized in that, The wavelet denoising method uses the db4 wavelet basis for 5-level decomposition. The neural network predicts the input of 15 historical sampling points and outputs the predicted value for the next 15 minutes.

Images