CN-122026347-A - Method, system, equipment, medium and product for controlling extreme event under typhoon

CN122026347ACN 122026347 ACN122026347 ACN 122026347ACN-122026347-A

Abstract

The application discloses a control method, a system, equipment, a medium and a product of an extreme event under typhoon, which are characterized by comprising the steps of obtaining typhoon weather data, structural parameters and real-time operation data of a power distribution network, mapping nodes and lines in the power distribution network into a space-time diagram structure, inputting the typhoon weather data and the real-time operation data into the space-time diagram structure to represent a dynamic coupling relation between typhoon evolution and power distribution network states to obtain coupling space-time characteristics, inputting the coupling space-time characteristics into a deep reinforcement learning control model to generate discrete action control instructions based on state vectors, generating corresponding continuous action control parameters according to the discrete action control instructions, determining a typhoon control strategy of the power distribution network by combining the discrete action control instructions and the continuous action control parameters, and carrying out collaborative optimization control on power distribution network equipment of the power distribution network under typhoon disasters based on the typhoon control strategy. The application can improve the accuracy of the cascade overload control of the power grid under typhoon disasters.

Inventors

TAN HUIJUAN
ZHAO RUIFENG
GUO WENXIN
LI QIAN
WANG CHEN
ZHANG SHUIPING
YI YANG

Assignees

广东电网有限责任公司电力调度控制中心

Dates

Publication Date: 20260512
Application Date: 20260122

Claims (10)

1. A method of controlling an extreme event in typhoons, comprising: Obtaining typhoon meteorological data, structural parameters of a power distribution network and real-time operation data; Mapping nodes and lines in the power distribution network into a space-time diagram structure based on topological connection relations in the structural parameters, and inputting typhoon meteorological data and real-time operation data into the space-time diagram structure to represent dynamic coupling relations between typhoon evolution and power distribution network states, so as to obtain coupling space-time characteristics for identifying weak links and fault evolution trends of the power distribution network; Inputting the coupling space-time characteristics into a preset deep reinforcement learning control model to generate a discrete action control instruction for controlling power distribution network equipment based on a state vector, generating corresponding continuous action control parameters according to the discrete action control instruction, and determining a typhoon control strategy of the power distribution network by combining the discrete action control instruction and the continuous action control parameters; and carrying out collaborative optimization control on power distribution network equipment of the power distribution network under typhoon disasters based on the typhoon control strategy.
2. The method for controlling extreme events under typhoons according to claim 1, wherein the mapping nodes and lines in the distribution network into a space-time diagram structure based on topological connection relations in the structural parameters comprises: Analyzing the structural parameters to obtain the node identification and electrical connection relation corresponding to each power element in the power distribution network; Determining a node set based on the node identification, and determining node characteristic vectors of all nodes in the node set; determining an edge set based on the electrical connection relation, and determining edge feature vectors of all circuits in the edge set; and constructing the space-time diagram structure based on the node set and the edge set.
3. The method for controlling an extreme event under typhoon according to claim 2, wherein the step of inputting the typhoon weather data and the real-time operation data into the space-time diagram structure to characterize a dynamic coupling relationship between typhoon evolution and a power distribution network state, and obtaining a coupling space-time feature for identifying a weak link and a fault evolution trend of the power distribution network comprises: Inputting the typhoon meteorological data and the real-time operation data into the space-time diagram structure to form a diagram state sequence evolving along with time; Performing convolution operation on the graph state sequence based on a time convolution network to obtain time correlation characteristics for representing typhoon evolution and power distribution network state change along with time; carrying out space feature aggregation on the time-associated features based on the graph rolling network and the electrical connection relation to obtain space features; And fusing the time correlation features with the space features to output coupling space-time features for identifying weak links and fault evolution trends of the power distribution network based on the fusion result.
4. The method of claim 1, wherein the inputting the coupled spatiotemporal features into a pre-set deep reinforcement learning control model to generate discrete motion control instructions for controlling power distribution network equipment based on state vectors comprises: inputting the coupling space-time characteristics, the preset random noise vectors and the preset physical constraint items into a generation countermeasure network so as to generate candidate power grid fault scenes through a generator; based on a discriminator, carrying out consistency discrimination on the candidate power grid fault scene and a historical real power grid fault scene to obtain a target power grid fault scene meeting typhoon disaster evolution characteristics, wherein the generating countermeasure network comprises the generator and the discriminator; Extracting physical quantity time sequence data used for representing the running state of power distribution network equipment from the target power grid fault scene, and processing the physical quantity time sequence data to obtain a state vector; And inputting the state vector into a discrete decision branch of the deep reinforcement learning control model to screen a plurality of candidate discrete actions corresponding to the power distribution network equipment under action mask constraint based on a preset greedy strategy, and determining a discrete action control instruction for controlling the power distribution network equipment.
5. The method of claim 1, wherein generating the corresponding continuous motion control parameters from the discrete motion control instructions comprises: inputting the discrete action control instruction into a continuous parameter branch of the deep reinforcement learning control model to generate corresponding initial continuous action control parameters by performing forward calculation on each discrete action; and carrying out constraint processing on the initial continuous action control parameters so as to limit the initial continuous action control parameters to the safety range of the power distribution network equipment and output the continuous action control parameters.
6. A method of controlling extreme events under typhoons according to claim 1, wherein said determining a typhoon control strategy for a power distribution network in combination with said discrete action control instructions and said continuous action control parameters comprises: combining the discrete motion control instruction and the continuous motion control parameter to form a hybrid control motion; Inputting the mixed control actions into a steady-state simulation environment to calculate the power flow of the power distribution network equipment under the disturbance of typhoon disasters, so as to obtain a system state; And carrying out strategy iterative optimization by using the deep reinforcement learning control model and taking the expected maximization of the accumulated value of the reward function as a target based on the system state, the mixed control action and a preset reward function until the generated multi-stage mixed control action sequence meets the preset condition, so as to obtain a typhoon control strategy of the power distribution network.
7. The control system for the extreme event under typhoon is characterized by comprising an acquisition module, a coupling module, a decision module and a control module; the acquisition module is used for acquiring typhoon meteorological data, structural parameters of the power distribution network and real-time operation data; The coupling module is used for mapping nodes and lines in the power distribution network into a space-time diagram structure based on topological connection relations in the structural parameters, inputting typhoon meteorological data and real-time operation data into the space-time diagram structure to represent dynamic coupling relations between typhoon evolution and power distribution network states, and obtaining coupling space-time characteristics for identifying weak links and fault evolution trends of the power distribution network; The decision module is used for inputting the coupling space-time characteristics into a preset deep reinforcement learning control model, generating a discrete action control instruction for controlling power distribution network equipment based on a state vector, generating corresponding continuous action control parameters according to the discrete action control instruction, and determining a typhoon control strategy of the power distribution network by combining the discrete action control instruction and the continuous action control parameters; And the control module is used for carrying out collaborative optimization control on power distribution network equipment of the power distribution network under typhoon disasters based on the typhoon control strategy.
8. A terminal device comprising a processor, a memory and a computer program stored in the memory and configured to be executed by the processor, the processor implementing the method of controlling a typhoon extreme event according to any one of claims 1-6 when executing the computer program.
9. A computer readable storage medium comprising a stored computer program, wherein the computer program, when run, controls a device in which the computer readable storage medium is located to perform the method of controlling an extreme event under typhoon according to any one of claims 1-6.
10. A computer program product comprising a computer program or instructions which, when executed by a communication device, implements a method of controlling an extreme event under typhoons according to any one of claims 1 to 6.

Description

Method, system, equipment, medium and product for controlling extreme event under typhoon Technical Field The invention relates to the field of emergency control of power grids, in particular to a control method, a system, equipment, a medium and a product for an extreme event in typhoons. Background Typhoons and other extreme natural disasters can cause transmission line breakage, power grid topological structure mutation and rapid load migration, so that local overload and cascading failure are caused, and safe and stable operation of a power grid is seriously threatened. Therefore, aiming at the rapid change of the running state of the power grid under the disaster condition, the power grid emergency control method is urgently needed to be developed, and the power flow balance and overload relief of the system are realized by timely adjusting the output of a unit, the load distribution and the adjustable equipment parameters, so that the safety and the reliability of the power grid under the extreme event are ensured. However, the existing power grid emergency control method has the problem of insufficient accuracy in coping with complex disaster scenes. The traditional method is dependent on static safety criteria or experience rules, rapid dynamic changes of the power grid state in the disaster process of typhoons and the like cannot be fully characterized, so that the identification of overload nodes and weak links is not accurate, and meanwhile, the response prediction capability for continuous adjustable control resources (such as phase shifter adjustment) and multistage disaster evolution is limited, so that overload cannot be accurately eliminated in the practical application of a control strategy, and the overall reliability of emergency control of the power grid is reduced. Disclosure of Invention The invention provides a control method, a system, equipment, a medium and a product for an extreme event in typhoons, which can improve the accuracy of cascade overload control of a power grid under typhoons. In a first aspect, an embodiment of the present invention provides a method for controlling an extreme event in typhoons, including: Obtaining typhoon meteorological data, structural parameters of a power distribution network and real-time operation data; Mapping nodes and lines in the power distribution network into a space-time diagram structure based on topological connection relations in the structural parameters, and inputting typhoon meteorological data and real-time operation data into the space-time diagram structure to represent dynamic coupling relations between typhoon evolution and power distribution network states, so as to obtain coupling space-time characteristics for identifying weak links and fault evolution trends of the power distribution network; Inputting the coupling space-time characteristics into a preset deep reinforcement learning control model to generate a discrete action control instruction for controlling power distribution network equipment based on a state vector, generating corresponding continuous action control parameters according to the discrete action control instruction, and determining a typhoon control strategy of the power distribution network by combining the discrete action control instruction and the continuous action control parameters; and carrying out collaborative optimization control on power distribution network equipment of the power distribution network under typhoon disasters based on the typhoon control strategy. The embodiment of the invention provides comprehensive and accurate data basis for analyzing the influence of typhoons on a power grid by collecting weather information such as wind speed, air pressure, air temperature and the like under the influence of typhoons and real-time states of the power grid topology, line parameters, node types and power generation/load, couples the dynamic influence of typhoons with the running state of the power grid by constructing a graph structure reflecting the node and line topology and time evolution of the power grid, can reveal the position where overload or instability easily occurs in key lines and nodes in the typhoons, provides accurate risk indexes for subsequent control strategies, converts the coupling characteristics into operable control decisions by utilizing a reinforcement learning model, realizes the joint optimization of discrete actions (power generator outage, load shedding and phase shifter adjustment) and continuous parameters (power generator shedding power, load shedding proportion and phase shifter phase adjustment angle), enables the control strategy to dynamically respond to typhoons in different stages, improves the accuracy and the performability of decision, guides the power flow to redistribute by accurately executing multi-period and mixed control actions, reduces the power flow, blocks the potential cascade propagation paths, and effectively inhibits the spread faults