Power distribution network elasticity improvement method based on distributed new energy aggregation

Abstract

A power distribution network elasticity improvement method based on distributed new energy aggregation relates to the technical field of power systems, and comprises the following steps: S1, establishing a device-cluster-system three-level collaborative architecture, and collecting operation data of distributed energy nodes in real time through digital twin dynamic modeling; s2, dynamically adjusting a virtual power plant aggregation boundary based on a deep reinforcement learning algorithm, and optimizing a resource response strategy; s3, a multi-time-scale optimization framework is adopted, and space-time cooperative elastic enhancement is achieved in the pre-disaster stage, the in-disaster stage and the post-disaster stage; s4, realizing credible evidence storage and compensation liquidation of the resource scheduling instruction through the block chain; aiming at the problems of dynamic response delay and insufficient space-time dimension collaboration in the existing distributed energy aggregation technology, the invention realizes full-chain elastic improvement from minute-level dynamic aggregation to hour-level recovery optimization by constructing a three-level collaboration architecture, and solves the problems of low resource utilization rate and slow after-disaster recovery speed in the traditional method.

Description

Technical Field

[0001] The present invention relates to the technical field of power systems, and in particular to a method for improving the elasticity of a distribution network based on distributed new energy aggregation. Background Art

[0002] When dealing with extreme natural disasters (such as typhoons and ice disasters), current distribution networks generally face two major technical bottlenecks: dynamic response hysteresis and global coordination failure. Traditional distributed energy aggregation technologies mostly use static resource pool division, which results in the rigidification of resource adjustment boundaries during disasters and is unable to adapt to dynamic fault scenarios. For example, the existing virtual power plant (VPP) aggregation model relies on fixed droop coefficients (such as K droop =1.0) and preset power adjustment range (±10%P N ), when a sudden impedance change (ΔZ > 10%) occurs in the backbone network, response delays exceed 30 seconds, and the critical load recovery rate is less than 60%. Furthermore, existing technologies lack a multi-dimensional spatiotemporal coordination mechanism, and pre-disaster pre-aggregation, dynamic adjustment during a disaster, and post-disaster recovery optimization are disconnected. Pre-disaster resource scheduling fails to consider meteorological data-driven line vulnerability assessments (e.g., calculation of the wind deviation coefficient γ is missing); mid-disaster blackstarts rely on a centralized control architecture (communication delay > 200ms); and post-disaster repair route planning ignores the integration of drone inspection data (damage identification accuracy <70%). These shortcomings result in insufficient distribution network resilience, unit load restoration costs as high as 2.5 yuan / kWh, and voltage compliance rates below 90%. Summary of the Invention

[0003] In order to overcome the deficiencies in the background technology, the present invention discloses a method for improving the elasticity of a distribution network based on distributed new energy aggregation.

[0004] In order to achieve the above-mentioned object of the invention, the present invention adopts the following technical solutions:

[0005] A method for improving the elasticity of a distribution network based on distributed renewable energy aggregation includes the following steps:

[0006] S1. Establish a three-level collaborative architecture of equipment, cluster, and system, and collect the operating data of distributed energy nodes in real time through digital twin dynamic modeling;

[0007] S2. Dynamically adjust virtual power plant (VPP) aggregation boundaries based on deep reinforcement learning algorithms to optimize resource response strategies;

[0008] S3. Adopt a multi-time-scale optimization framework to achieve spatiotemporal resilience enhancement in the pre-disaster, during-disaster, and post-disaster stages;

[0009] S4. Realize the trusted evidence storage and compensation settlement of resource scheduling instructions through blockchain.

[0010] Preferably, the digital twin dynamic modeling includes:

[0011] Deploy edge intelligent terminals to collect real-time data on photovoltaic IV curves, energy storage SOC status, and flexible load demand, with a sampling frequency of ≥1kHz;

[0012] Based on the OPC UA protocol, multi-source data is standardized and mapped to build a dynamic twin database with an update cycle of ≤5 seconds.

[0013] Preferably, the deep reinforcement learning algorithm is a DDPG algorithm, whose state space includes node voltage deviation, energy storage capacity and load priority weight, the action space is defined as the VPP capacity adjustment range ±20% rated power and the droop control coefficient 0.8-1.2, and the reward function comprehensively considers voltage stability, critical load recovery rate and operating cost.

[0014] Preferably, the multi-timescale optimization framework includes:

[0015] Pre-aggregation 24 hours before a disaster: The NSGA-II algorithm is used to generate the Pareto optimal solution set, with the objective function being to minimize the power purchase cost and maximize voltage stability;

[0016] 5-minute dynamic adjustment during a disaster: When an impedance mutation ΔZ > 15% is detected, a black start is triggered, the energy storage system switches to VSG mode, and the topology is optimized.

[0017] Optimize one-hour post-disaster recovery: Combine drone inspection data to establish a joint model of load recovery rate and repair time, and use the branch and bound method to solve the optimal path.

[0018] Preferably, the blockchain includes:

[0019] Based on Hyperledger Fabric, timestamps, scheduling instructions and load status are recorded; when the actual recovery amount is lower than 90% of the promised value, the smart contract automatically triggers a 1.5 times difference in electricity charges.

[0020] Due to the adoption of the above-mentioned technical solution, the present invention has the following beneficial effects:

[0021] The present invention discloses a method for improving the elasticity of distribution networks based on distributed renewable energy aggregation. The VPP dynamic aggregation model (DDPG algorithm) based on deep reinforcement learning can extend the resource regulation boundary to ±20%P N , droop coefficient K droopThe dynamic adjustment range reaches [0.5, 2.0], shortening disaster response time to less than 2 seconds and increasing the critical load recovery rate to over 89%. A line vulnerability assessment model integrating meteorological data (wind deviation coefficient γ = 0.6V_wind² / (g·d)) pre-generates 12 dynamic VPPs with an adjustable capacity of 58MW, improving pre-disaster resource utilization by 35%.

[0022] Before the disaster, an improved NSGA-II algorithm was used to generate a Pareto frontier solution, and the comprehensive optimization rate of electricity purchase costs and operation and maintenance costs increased by 28%. During the disaster, VSG control was used to achieve rapid microgrid networking (island establishment time <5 seconds), and the voltage qualification rate increased from 91.5% to 98.7%.

[0023] After the disaster, the ADMM coordination mechanism was integrated with the drone damage heat map to generate the optimal repair path, shortening the total recovery time from 4.2 hours to 2.3 hours and reducing the unit load recovery cost to 1.8 yuan / kWh; a three-dimensional resilience assessment system (static / dynamic / economic resilience) was established, and the comprehensive resilience index R index ≥0.85, providing a quantitative basis for optimizing the elasticity of the distribution network. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a flowchart of the present invention. DETAILED DESCRIPTION

[0025] The present invention can be explained in detail through the following embodiments. The purpose of disclosing the present invention is to protect all technical improvements within the scope of the present invention. In the description of the present invention, it should be understood that if there are terms such as "upper", "lower", "front", "back", "left", "right", etc. to indicate directions or positional relationships, they only correspond to the drawings of this application for the convenience of describing the present invention, and do not indicate or imply that the device or element referred to must have a specific direction.

[0026] Example 1, combined with the attached Figure 1 A method for improving the elasticity of a distribution network based on distributed renewable energy aggregation includes the following steps:

[0027] S1. Establish a three-level collaborative architecture of equipment, cluster and system, and collect the operating data of distributed energy nodes in real time through digital twin dynamic modeling.

[0028] Specifically, edge intelligent terminals (ARM Cortex-A72 processors + 5G communication modules) are deployed at each distributed energy node to collect real-time data such as PV string IV curves (sampling frequency 1kHz), energy storage SOC status (accuracy ±0.5%), and flexible load demand. Multi-source data is standardized and mapped using the OPC UA protocol. A dynamic twin database is established, encompassing device physical parameters (such as inverter efficiency curves η = 0.95-0.98) and environmental sensing data (temperature, humidity, and irradiance), with a data update cycle of ≤5s.

[0029] S2. Dynamically adjust the virtual power plant (VPP) aggregation boundaries based on deep reinforcement learning algorithms to optimize resource response strategies.

[0030] Specifically, the virtual power plant (VPP) boundary adjustment mechanism based on the DDPG algorithm includes:

[0031] State space: 36-dimensional vector, including node voltage deviation ΔU (allowable range ±7%), energy storage remaining capacity C ESS (0~100%), load priority weight ω Z (Divided into 3 levels: );

[0032] Action space: defines the VPP aggregation capacity adjustment range ±20%P N (rated power), droop control coefficient K P (0.8~1.2);

[0033] Reward function: ,in , , , , Ten thousand yuan.

[0034] S3. Adopt a multi-time-scale optimization framework to achieve spatiotemporal coordinated resilience enhancement in the three stages of pre-disaster, during disaster, and post-disaster.

[0035] Specifically, pre-disaster aggregation (24h scale):

[0036] The improved NSGA-II algorithm is used to generate the Pareto optimal solution set, and the objective function is:

[0037] in, is the electricity purchase cost during period t (yuan / kWh), The cost of distributed power generation operation and maintenance (including depreciation of 0.15 yuan / kWh), It is the voltage stability index of key nodes (threshold ≥ 0.9).

[0038] Dynamic adjustment during disaster (5-minute scale):

[0039] Deploy a distributed optimization controller based on a consensus algorithm. When a backbone network fault is detected (impedance mutation ΔZ > 15%), a black start sequence is triggered:

[0040] Energy storage switches to VSG mode to provide voltage support (U ref =1.0) and frequency adjustment (f ref =50±0.2Hz);

[0041] The photovoltaic inverter enables reactive power priority control, and the Q output capability is increased to ±60%S N (rated capacity);

[0042] Generate a minimum tree topology based on Prim's algorithm (weight coefficient β = 0.7).

[0043] Post-disaster recovery optimization (1h scale):

[0044] Combined with the line damage heat map (resolution 0.5m) generated by drone inspection, a joint optimization model was established:

[0045]

[0046] in, is the load recovery rate (target ≥95%), is the repair time (unit: h), λ=0.3 is the weight coefficient, and the optimal path is solved by the branch and bound method.

[0047] Double-layer spatial domain collaboration:

[0048] Feeder layer: Dynamically divide power supply zones based on impedance spectrum analysis (frequency range 50-1500Hz). When the feeder load rate is >85%, the adjacent VPP power support (maximum support amount ΔP max =30%P N );

[0049] User layer: Deploy intelligent measurement terminals (accuracy level 0.5S) to identify load characteristics. The temperature control load regulation potential is modeled as follows:

[0050]

[0051] Among them is is 0.8 (adjustable coefficient), ±2℃ (allowable deviation), 3.0 (refrigeration efficiency)

[0052] S4. Realize the trusted evidence storage and compensation settlement of resource scheduling instructions through blockchain.

[0053] Specifically, the QV characteristic curve of the photovoltaic inverter is discretized into a 10-segment linear model (fitting error ≤ 1%); a PI controller (K P =1.2, K i =0.05) to implement VPP instruction decomposition.

[0054] Constructing three-dimensional evaluation indicators:

[0055] Static toughness: spare capacity ratio ≥ 20%;

[0056] Dynamic resilience: Fault recovery rate ≥ 5MW / min;

[0057] Economic resilience: Unit load recovery cost ≤ 2 yuan / kWh.

[0058] Calculation of comprehensive elasticity index R by entropy weight-TOPSIS method index (Target value ≥ 0.85).

[0059] A blockchain based on Hyperledger Fabric was built, and a smart contract was designed. This records timestamps (UTC format), resource scheduling instructions (JSON format), and load recovery status (0 / 1 flag). When the actual recovery amount is less than 90% of the promised value, compensation is automatically triggered (compensation rate γ = 1.5 times the difference in electricity charges).

[0060] Take a coastal area (average annual typhoons ≥ 3 times) as an example:

[0061] Deployed edge intelligent terminals (model ET-3260) equipped with quad-core ARM processors and 5G modules at 23 photovoltaic power plants (single-site capacity ≥ 2MW);

[0062] 18 energy storage systems (500kWh / 250kW per system) were installed, equipped with VSG control modules (response time <100ms);

[0063] Intelligent measurement terminals (sampling period 1s) are configured for 56 key loads (hospitals, communication base stations).

[0064] The digital twin platform uses the OpenDSS kernel with a simulation step of 1s;

[0065] The DDPG algorithm is implemented in TensorFlow 2.8, with a training cycle of 72 hours (NVIDIA A100 GPU).

[0066] The blockchain nodes are deployed on 4 servers (CPU: Intel Xeon Gold 6338, memory: 256GB).

[0067] 24 hours before the disaster: Run the NSGA-II algorithm to generate a pre-aggregation plan, which is published to each VPP via OPC UA.

[0068] 5 minutes into the disaster: The consistency algorithm adjusts the power mutual aid in real time, and the islanding detection module refreshes its status every 2 seconds.

[0069] 1 hour after the disaster: UAV inspection data is called (updated every 10 minutes), and the ADMM algorithm optimizes the repair path every 30 minutes.

[0070] It has been verified that the average load recovery time has been shortened from 4.2h to 2.3h (an increase of 45.2%); the voltage qualification rate at key nodes has been increased from 91.5% to 98.7%.

[0071] The parts of the present invention that are not described in detail are prior art. It is obvious to those skilled in the art that the present invention is not limited to the details of the above-mentioned exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or basic characteristics of the present invention. Therefore, no matter from which point of view, the embodiments should be regarded as exemplary and non-restrictive, and it is intended that all changes that fall within the meaning and scope of equivalent elements are included in the present invention.

Claims

1. A method for improving the elasticity of distribution networks based on distributed renewable energy aggregation, characterized by: The following steps are involved: S1. Establish a three-level collaborative architecture of equipment, cluster, and system, and collect the operating data of distributed energy nodes in real time through digital twin dynamic modeling; S2. Dynamically adjust virtual power plant (VPP) aggregation boundaries based on deep reinforcement learning algorithms to optimize resource response strategies; S3. Adopt a multi-time-scale optimization framework to achieve spatiotemporal resilience enhancement in the pre-disaster, during-disaster, and post-disaster stages; S4. Realize the trusted evidence storage and compensation settlement of resource scheduling instructions through blockchain.

2. The method for improving the elasticity of distribution networks based on distributed renewable energy aggregation according to claim 1, characterized in that: The digital twin dynamic modeling includes: Deploy edge intelligent terminals to collect real-time data on photovoltaic IV curves, energy storage SOC status, and flexible load demand, with a sampling frequency of ≥1kHz; Based on the OPC UA protocol, multi-source data is standardized and mapped to build a dynamic twin database with an update cycle of ≤5 seconds.

3. The method for improving the elasticity of distribution networks based on distributed renewable energy aggregation according to claim 1, characterized in that: The deep reinforcement learning algorithm is the DDPG algorithm, whose state space includes node voltage deviation, energy storage capacity and load priority weight. The action space is defined as the VPP capacity adjustment range ±20% rated power and the droop control coefficient of 0.8-1.

2. The reward function integrates voltage stability, critical load recovery rate and operating cost.

4. The method for improving the elasticity of a distribution network based on distributed renewable energy aggregation according to claim 1, characterized in that: The multi-timescale optimization framework includes: Pre-aggregation 24 hours before a disaster: The NSGA-II algorithm is used to generate the Pareto optimal solution set, with the objective function being to minimize the power purchase cost and maximize voltage stability; 5-minute dynamic adjustment during a disaster: When an impedance mutation ΔZ > 15% is detected, a black start is triggered, the energy storage system switches to VSG mode, and the topology is optimized. Optimize one-hour post-disaster recovery: Combine drone inspection data to establish a joint model of load recovery rate and repair time, and use the branch and bound method to solve the optimal path.

5. The method for improving the elasticity of distribution networks based on distributed renewable energy aggregation according to claim 1, characterized in that: The blockchain includes: Based on Hyperledger Fabric, timestamps, scheduling instructions and load status are recorded; when the actual recovery amount is lower than 90% of the promised value, the smart contract automatically triggers a 1.5 times difference in electricity charges.

Images